Spaces:
Paused
Paused
Add Gradio frontend: 5-tab UI with full demo loop screens
Browse files- Build complete Gradio UI (Explore, Bookshelf, Clone Voice Studio, Companion Player, Settings)
- Implement Ask/Resume/Finish Q&A flow in Player tab (demo loop steps 4-8)
- Add procedural chime audio generation as audio placeholder
- Wire state callbacks: book mounting, shelf filter, voice cloning pipeline, subtitle sync, error sandbox
- Q&A panel: text + mic input, answer display, answer audio output widget
- Story-finished prompt triggered at timeline end; Resume restores PLAYING state
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
app.py
ADDED
|
@@ -0,0 +1,1080 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import math
|
| 3 |
+
import struct
|
| 4 |
+
import wave
|
| 5 |
+
import gradio as gr
|
| 6 |
+
import time
|
| 7 |
+
|
| 8 |
+
# Create directories for sample audio files
|
| 9 |
+
os.makedirs("sample_sounds", exist_ok=True)
|
| 10 |
+
|
| 11 |
+
# Generate relaxing audio chime waveforms procedurally
|
| 12 |
+
# This runs fully client/server-side with standard Python, eliminating mock limits or Torch wait times
|
| 13 |
+
def generate_chimes_wav(filename, duration=120, melody_type="lullaby"):
|
| 14 |
+
sample_rate = 16000
|
| 15 |
+
n_samples = int(duration * sample_rate)
|
| 16 |
+
|
| 17 |
+
with wave.open(filename, 'wb') as wav_file:
|
| 18 |
+
wav_file.setnchannels(1)
|
| 19 |
+
wav_file.setsampwidth(2)
|
| 20 |
+
wav_file.setframerate(sample_rate)
|
| 21 |
+
|
| 22 |
+
# Pentatonic cozy scales
|
| 23 |
+
if melody_type == "lullaby":
|
| 24 |
+
notes = [261.63, 293.66, 329.63, 392.00, 440.00] # C4, D4, E4, G4, A4
|
| 25 |
+
elif melody_type == "adventure":
|
| 26 |
+
notes = [196.00, 246.94, 293.63, 329.63, 392.00] # G3, B3, D4, E4, G4
|
| 27 |
+
else:
|
| 28 |
+
notes = [220.00, 261.63, 293.66, 329.63, 392.00] # A3, C4, D4, E4, G4
|
| 29 |
+
|
| 30 |
+
for i in range(n_samples):
|
| 31 |
+
# cycle notes based on melody speed
|
| 32 |
+
note_speed = sample_rate * 1.5 if melody_type == "lullaby" else sample_rate * 1.2
|
| 33 |
+
note_idx = int((i / note_speed) % len(notes))
|
| 34 |
+
freq = notes[note_idx]
|
| 35 |
+
|
| 36 |
+
# envelope to prevent crackles
|
| 37 |
+
sample_in_note = i % int(note_speed)
|
| 38 |
+
envelope = 1.0
|
| 39 |
+
if sample_in_note < 1200: # attack
|
| 40 |
+
envelope = sample_in_note / 1200
|
| 41 |
+
else: # decay
|
| 42 |
+
decay_length = note_speed - 1200
|
| 43 |
+
envelope = max(0.0, 1.0 - (sample_in_note - 1200) / decay_length)
|
| 44 |
+
|
| 45 |
+
# Synthesize voice-harmonic chime
|
| 46 |
+
val = math.sin(2 * math.pi * freq * i / sample_rate)
|
| 47 |
+
val += 0.45 * math.sin(2 * math.pi * (freq * 1.5) * i / sample_rate)
|
| 48 |
+
val += 0.25 * math.sin(2 * math.pi * (freq * 2.0) * i / sample_rate)
|
| 49 |
+
val = val / 1.7 * envelope
|
| 50 |
+
|
| 51 |
+
packed_val = struct.pack('<h', int(val * 16384))
|
| 52 |
+
wav_file.writeframes(packed_val)
|
| 53 |
+
|
| 54 |
+
# Instantiate the audio files immediately so they are hot in HF Spaces cache
|
| 55 |
+
create_sound_library = {
|
| 56 |
+
"willow": ("sample_sounds/willow.wav", 120, "lullaby"),
|
| 57 |
+
"sky": ("sample_sounds/sky.wav", 180, "adventure"),
|
| 58 |
+
"deep": ("sample_sounds/deep.wav", 240, "adventure"),
|
| 59 |
+
"baker": ("sample_sounds/baker.wav", 98, "lullaby"),
|
| 60 |
+
"letters": ("sample_sounds/letters.wav", 156, "classic"),
|
| 61 |
+
"cloned_preview": ("sample_sounds/cloned_preview.wav", 15, "lullaby")
|
| 62 |
+
}
|
| 63 |
+
|
| 64 |
+
for key, (path, dur, mel) in create_sound_library.items():
|
| 65 |
+
if not os.path.exists(path):
|
| 66 |
+
generate_chimes_wav(path, duration=dur, melody_type=mel)
|
| 67 |
+
|
| 68 |
+
# Initial dataset of audiobooks
|
| 69 |
+
mock_books = [
|
| 70 |
+
{
|
| 71 |
+
"id": "1",
|
| 72 |
+
"title": "The Enchanted Willow",
|
| 73 |
+
"author": "Elena S. Thorne",
|
| 74 |
+
"cover_url": "https://images.unsplash.com/photo-1544947950-fa07a98d237f?auto=format&fit=crop&q=80&w=400",
|
| 75 |
+
"voice_name": "Mom's Voice",
|
| 76 |
+
"duration": 1200,
|
| 77 |
+
"elapsed_time": 456,
|
| 78 |
+
"synopsis": "Deep within the heart of Whispering Woods stands an ancient willow tree, rumored to hold the forgotten memories of the valley. When a young cartographer is sent to map the region, she starts hearing a gentle, familiar voice whispering from the rustling emerald leaves.",
|
| 79 |
+
"category": "Fantasy",
|
| 80 |
+
"is_cloned": True,
|
| 81 |
+
"percentage": 38,
|
| 82 |
+
"audio_path": "sample_sounds/willow.wav"
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"id": "2",
|
| 86 |
+
"title": "The Clockwork Sky",
|
| 87 |
+
"author": "Arthur Pendelton",
|
| 88 |
+
"cover_url": "https://images.unsplash.com/photo-1512820790803-83ca734da794?auto=format&fit=crop&q=80&w=400",
|
| 89 |
+
"voice_name": "Folk Storyteller",
|
| 90 |
+
"duration": 1800,
|
| 91 |
+
"elapsed_time": 0,
|
| 92 |
+
"synopsis": "In a world where the heavens are driven by complex brass gears and clockwork stars, a young apprentice discovers an unaccounted gear grinding in the midnight skyβa secret that the High Astrologers have hidden for a millennium.",
|
| 93 |
+
"category": "Sci-Fi & Steampunk",
|
| 94 |
+
"is_cloned": False,
|
| 95 |
+
"percentage": 0,
|
| 96 |
+
"audio_path": "sample_sounds/sky.wav"
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"id": "3",
|
| 100 |
+
"title": "Echoes of the Deep",
|
| 101 |
+
"author": "Mariana Vance",
|
| 102 |
+
"cover_url": "https://images.unsplash.com/photo-1476275466078-4007374efbbe?auto=format&fit=crop&q=80&w=400",
|
| 103 |
+
"voice_name": "Grandpa Joseph",
|
| 104 |
+
"duration": 2400,
|
| 105 |
+
"elapsed_time": 1200,
|
| 106 |
+
"synopsis": "A quiet oceanographer working in a remote marine sanctuary begins to trace an unusual, rhythmic hum coming from the Marianas Trench. The signal grows stronger, mimicking a lullaby his grandfather used to whistle.",
|
| 107 |
+
"category": "Adventure",
|
| 108 |
+
"is_cloned": True,
|
| 109 |
+
"percentage": 50,
|
| 110 |
+
"audio_path": "sample_sounds/deep.wav"
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"id": "4",
|
| 114 |
+
"title": "The Midnight Baker",
|
| 115 |
+
"author": "Clara Dupont",
|
| 116 |
+
"cover_url": "https://images.unsplash.com/photo-1456513080510-7bf3a84b82f8?auto=format&fit=crop&q=80&w=400",
|
| 117 |
+
"voice_name": "Mom's Voice",
|
| 118 |
+
"duration": 980,
|
| 119 |
+
"elapsed_time": 980,
|
| 120 |
+
"synopsis": "Every night at midnight, Clara bakes bread that infuses the sleeping town with warmth. Customers donβt buy her pastries to satisfy hunger; they eat them to relive the comforting memories of childhood summers.",
|
| 121 |
+
"category": "Slice of Life",
|
| 122 |
+
"is_cloned": True,
|
| 123 |
+
"percentage": 100,
|
| 124 |
+
"audio_path": "sample_sounds/baker.wav"
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"id": "5",
|
| 128 |
+
"title": "Letters from Paris",
|
| 129 |
+
"author": "Julian Mercer",
|
| 130 |
+
"cover_url": "https://images.unsplash.com/photo-1492052722242-2554d0e99e3a?auto=format&fit=crop&q=80&w=400",
|
| 131 |
+
"voice_name": "Grandpa Joseph",
|
| 132 |
+
"duration": 1560,
|
| 133 |
+
"elapsed_time": 0,
|
| 134 |
+
"synopsis": "A bundle of yellowed, ribbon-bound letters hidden inside an attic chest on Rue de lβOdeon unravels a heartbreaking secret of courage and hidden artwork dating back to the winter of 1944.",
|
| 135 |
+
"category": "Historical Fiction",
|
| 136 |
+
"is_cloned": True,
|
| 137 |
+
"percentage": 0,
|
| 138 |
+
"audio_path": "sample_sounds/letters.wav"
|
| 139 |
+
}
|
| 140 |
+
]
|
| 141 |
+
|
| 142 |
+
mock_voices = [
|
| 143 |
+
{
|
| 144 |
+
"id": "v1",
|
| 145 |
+
"name": "Mom's Voice",
|
| 146 |
+
"avatar_url": "π©βπ¦³",
|
| 147 |
+
"description": "Warm, soft, with a natural soothing cadence. Best for bedtime stories.",
|
| 148 |
+
"gender": "Female",
|
| 149 |
+
"created_date": "2026-06-01",
|
| 150 |
+
"status": "ready"
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"id": "v2",
|
| 154 |
+
"name": "Grandpa Joseph",
|
| 155 |
+
"avatar_url": "π΄",
|
| 156 |
+
"description": "Deep, resonant, with an occasional comforting chuckle. Perfect for classic adventure stories.",
|
| 157 |
+
"gender": "Male",
|
| 158 |
+
"created_date": "2026-06-03",
|
| 159 |
+
"status": "ready"
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"id": "v3",
|
| 163 |
+
"name": "Aunt Sarah",
|
| 164 |
+
"avatar_url": "π©",
|
| 165 |
+
"description": "Bubbly, expressive, brilliant at acting out funny character voices.",
|
| 166 |
+
"gender": "Female",
|
| 167 |
+
"created_date": "2026-06-06",
|
| 168 |
+
"status": "processing"
|
| 169 |
+
}
|
| 170 |
+
]
|
| 171 |
+
|
| 172 |
+
narrative_subtitles = [
|
| 173 |
+
"Deep within the heart of Whispering Woods stands an ancient willow tree...",
|
| 174 |
+
"It was rumored to hold the forgotten memories of the valley.",
|
| 175 |
+
"When a young cartographer is sent to map the region...",
|
| 176 |
+
"She starts hearing a gentle, familiar voice whispering from the emerald leaves.",
|
| 177 |
+
"The voice seemed to know her name, calling softly in the early autumn breeze...",
|
| 178 |
+
"And that was when she realized, she was not alone in the valley."
|
| 179 |
+
]
|
| 180 |
+
|
| 181 |
+
css_code = """
|
| 182 |
+
@import url('https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=Playfair+Display:ital,wght@0,400;0,600;0,700;1,400&display=swap');
|
| 183 |
+
/* Main layouts and colors matching VoiceBook cream and amber theme */
|
| 184 |
+
body, .gradio-container {
|
| 185 |
+
background-color: #FAF7F2 !important;
|
| 186 |
+
font-family: 'Inter', system-ui, sans-serif !important;
|
| 187 |
+
}
|
| 188 |
+
.gradio-container {
|
| 189 |
+
max-width: 1200px !important;
|
| 190 |
+
margin: 0 auto !important;
|
| 191 |
+
}
|
| 192 |
+
/* Serif headers */
|
| 193 |
+
.serif-header {
|
| 194 |
+
font-family: 'Playfair Display', Georgia, serif !important;
|
| 195 |
+
color: #1c1c19 !important;
|
| 196 |
+
font-weight: 700 !important;
|
| 197 |
+
}
|
| 198 |
+
.cream-bg {
|
| 199 |
+
background-color: #FAF7F2 !important;
|
| 200 |
+
}
|
| 201 |
+
.card-container {
|
| 202 |
+
background-color: #f6f3ee !important;
|
| 203 |
+
border: 1px solid #ebdccb !important;
|
| 204 |
+
border-radius: 16px !important;
|
| 205 |
+
padding: 24px !important;
|
| 206 |
+
box-shadow: 0 2px 8px rgba(148, 74, 0, 0.03) !important;
|
| 207 |
+
}
|
| 208 |
+
.accent-border {
|
| 209 |
+
border-color: #ddc1b0 !important;
|
| 210 |
+
}
|
| 211 |
+
.highlight-orange {
|
| 212 |
+
color: #f5841f !important;
|
| 213 |
+
}
|
| 214 |
+
.highlight-amber {
|
| 215 |
+
color: #944a00 !important;
|
| 216 |
+
}
|
| 217 |
+
.dark-card-container {
|
| 218 |
+
background-color: #122116 !important;
|
| 219 |
+
color: #FAF7F2 !important;
|
| 220 |
+
border: 1px solid #1e3422 !important;
|
| 221 |
+
border-radius: 16px !important;
|
| 222 |
+
padding: 24px !important;
|
| 223 |
+
box-shadow: 0 2px 8px rgba(0, 0, 0, 0.15) !important;
|
| 224 |
+
}
|
| 225 |
+
/* Customized HTML grids matching identical React typography */
|
| 226 |
+
.hero-spotlight {
|
| 227 |
+
background: linear-gradient(135deg, #1e3422 0%, #121f15 100%) !important;
|
| 228 |
+
color: #FFFFFF !important;
|
| 229 |
+
border-radius: 24px !important;
|
| 230 |
+
padding: 32px !important;
|
| 231 |
+
position: relative;
|
| 232 |
+
overflow: hidden;
|
| 233 |
+
}
|
| 234 |
+
.badge-featured {
|
| 235 |
+
background: rgba(245, 132, 31, 0.2);
|
| 236 |
+
color: #f5841f;
|
| 237 |
+
border: 1px solid rgba(245, 132, 31, 0.3);
|
| 238 |
+
border-radius: 9999px;
|
| 239 |
+
padding: 4px 10px;
|
| 240 |
+
font-size: 11px;
|
| 241 |
+
font-weight: 700;
|
| 242 |
+
text-transform: uppercase;
|
| 243 |
+
display: inline-block;
|
| 244 |
+
}
|
| 245 |
+
.book-shelf-grid {
|
| 246 |
+
display: grid;
|
| 247 |
+
grid-template-columns: repeat(auto-fill, minmax(280px, 1fr));
|
| 248 |
+
gap: 20px;
|
| 249 |
+
}
|
| 250 |
+
.shelf-card {
|
| 251 |
+
background: #f6f3ee;
|
| 252 |
+
border: 1px solid #eadacc;
|
| 253 |
+
border-radius: 16px;
|
| 254 |
+
padding: 16px;
|
| 255 |
+
display: flex;
|
| 256 |
+
gap: 16px;
|
| 257 |
+
transition: all 0.3s ease;
|
| 258 |
+
cursor: pointer;
|
| 259 |
+
}
|
| 260 |
+
.shelf-card:hover {
|
| 261 |
+
border-color: #f5841f;
|
| 262 |
+
transform: translateY(-2px);
|
| 263 |
+
box-shadow: 0 4px 12px rgba(148, 74, 0, 0.06);
|
| 264 |
+
}
|
| 265 |
+
.cover-image {
|
| 266 |
+
width: 80px;
|
| 267 |
+
height: 110px;
|
| 268 |
+
object-fit: cover;
|
| 269 |
+
border-radius: 8px;
|
| 270 |
+
box-shadow: 0 2px 6px rgba(0,0,0,0.1);
|
| 271 |
+
}
|
| 272 |
+
.progress-bar-bg {
|
| 273 |
+
width: 100%;
|
| 274 |
+
height: 6px;
|
| 275 |
+
background: #ebdccb;
|
| 276 |
+
border-radius: 999px;
|
| 277 |
+
overflow: hidden;
|
| 278 |
+
}
|
| 279 |
+
.progress-bar-fill {
|
| 280 |
+
height: 100%;
|
| 281 |
+
background: #f5841f;
|
| 282 |
+
}
|
| 283 |
+
/* Pulsing rings for microphone recording */
|
| 284 |
+
@keyframes pulse-ring {
|
| 285 |
+
0% { transform: scale(0.85); opacity: 0.6; }
|
| 286 |
+
50% { transform: scale(1.15); opacity: 0.25; }
|
| 287 |
+
100% { transform: scale(1.4); opacity: 0; }
|
| 288 |
+
}
|
| 289 |
+
.pulse-circle {
|
| 290 |
+
position: relative;
|
| 291 |
+
width: 120px;
|
| 292 |
+
height: 120px;
|
| 293 |
+
border-radius: 50%;
|
| 294 |
+
background: #944a00;
|
| 295 |
+
color: white;
|
| 296 |
+
display: flex;
|
| 297 |
+
flex-direction: column;
|
| 298 |
+
align-items: center;
|
| 299 |
+
justify-content: center;
|
| 300 |
+
margin: 0 auto;
|
| 301 |
+
}
|
| 302 |
+
.recording-active {
|
| 303 |
+
background: #ef4444 !important;
|
| 304 |
+
}
|
| 305 |
+
/* Subtitle dialog ticker */
|
| 306 |
+
.dialog-synced {
|
| 307 |
+
background: rgba(0,0,0,0.15);
|
| 308 |
+
border: 1px solid rgba(255,166,0,0.1);
|
| 309 |
+
color: #ffd3a6;
|
| 310 |
+
font-family: 'Playfair Display', Georgia, serif;
|
| 311 |
+
font-size: 1.25rem;
|
| 312 |
+
font-style: italic;
|
| 313 |
+
padding: 24px;
|
| 314 |
+
border-radius: 16px;
|
| 315 |
+
text-align: center;
|
| 316 |
+
min-height: 100px;
|
| 317 |
+
display: flex;
|
| 318 |
+
align-items: center;
|
| 319 |
+
justify-content: center;
|
| 320 |
+
}
|
| 321 |
+
/* Tab button custom styles */
|
| 322 |
+
.tabs .tab-nav button {
|
| 323 |
+
border-bottom: 2px solid transparent !important;
|
| 324 |
+
}
|
| 325 |
+
.tabs .tab-nav button.selected {
|
| 326 |
+
color: #944a00 !important;
|
| 327 |
+
border-bottom-color: #944a00 !important;
|
| 328 |
+
font-weight: 700 !important;
|
| 329 |
+
}
|
| 330 |
+
"""
|
| 331 |
+
|
| 332 |
+
def generate_dashboard_html(books_list):
|
| 333 |
+
featured = books_list[0]
|
| 334 |
+
others = books_list[1:]
|
| 335 |
+
|
| 336 |
+
html = f"""
|
| 337 |
+
<div style="margin-bottom: 24px;">
|
| 338 |
+
<span style="font-size: 11px; text-transform: uppercase; letter-spacing: 2px; color: #f5841f; font-weight: 600;">Welcome Back, Sarah</span>
|
| 339 |
+
<h2 class="serif-header" style="font-size: 28px; margin-top: 4px; margin-bottom: 16px;">Explore Your Warm Audiobook Shelves</h2>
|
| 340 |
+
</div>
|
| 341 |
+
|
| 342 |
+
<!-- Hero Spotlight Section -->
|
| 343 |
+
<div class="hero-spotlight" style="display: flex; flex-direction: row; gap: 32px; align-items: center; margin-bottom: 32px; flex-wrap: wrap;">
|
| 344 |
+
<div style="flex: 1; min-width: 200px; max-width: 280px; text-align: center;">
|
| 345 |
+
<img src="{featured['cover_url']}" style="width: 180px; height: 260px; object-fit: cover; border-radius: 16px; box-shadow: 0 8px 24px rgba(0,0,0,0.3); transform: rotate(-2deg);" referrerPolicy="no-referrer" />
|
| 346 |
+
</div>
|
| 347 |
+
<div style="flex: 2; min-width: 300px; display: flex; flex-direction: column; justify-content: space-between; gap: 16px;">
|
| 348 |
+
<div>
|
| 349 |
+
<span class="badge-featured" style="margin-bottom: 8px;">Featured Clone</span>
|
| 350 |
+
<h3 class="serif-header" style="color: white !important; font-size: 28px; margin: 4px 0;">{featured['title']}</h3>
|
| 351 |
+
<p style="color: #6f6257;">by <span style="font-style: italic; color: #ddc1b0;">{featured['author']}</span></p>
|
| 352 |
+
<p style="color: #cbd5e1; font-size: 13.5px; line-height: 1.6; margin-top: 8px;">{featured['synopsis']}</p>
|
| 353 |
+
</div>
|
| 354 |
+
|
| 355 |
+
<div style="display: flex; gap: 12px; flex-wrap: wrap; font-size: 12px; color: #e2e8f0;">
|
| 356 |
+
<span style="background: rgba(255,255,255,0.1); padding: 4px 10px; border-radius: 8px;">β±οΈ 20 Min Remaining</span>
|
| 357 |
+
<span style="background: rgba(255,255,255,0.1); padding: 4px 10px; border-radius: 8px; font-weight: 600; color: #ffd3a6;">ποΈ Narrator: {featured['voice_name']}</span>
|
| 358 |
+
<span style="background: rgba(255,255,255,0.1); padding: 4px 10px; border-radius: 8px;">β Editorial Spotlight</span>
|
| 359 |
+
</div>
|
| 360 |
+
</div>
|
| 361 |
+
</div>
|
| 362 |
+
|
| 363 |
+
<div>
|
| 364 |
+
<h4 class="serif-header" style="font-size: 20px; margin-bottom: 16px;">Continue Reading with Channeled Voices</h4>
|
| 365 |
+
<div class="book-shelf-grid">
|
| 366 |
+
"""
|
| 367 |
+
|
| 368 |
+
for bk in books_list:
|
| 369 |
+
html += f"""
|
| 370 |
+
<div class="shelf-card">
|
| 371 |
+
<div>
|
| 372 |
+
<img src="{bk['cover_url']}" class="cover-image" referrerPolicy="no-referrer" />
|
| 373 |
+
</div>
|
| 374 |
+
<div style="display: flex; flex-direction: column; justify-content: space-between; flex: 1; min-width: 0;">
|
| 375 |
+
<div>
|
| 376 |
+
<div style="display: flex; justify-content: space-between; font-size: 10px; font-weight: 700; color: #f5841f;">
|
| 377 |
+
<span>{bk['category']}</span>
|
| 378 |
+
<span>{bk['percentage']}%</span>
|
| 379 |
+
</div>
|
| 380 |
+
<h5 class="serif-header" style="font-size: 14px; margin: 4px 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap;">{bk['title']}</h5>
|
| 381 |
+
<p style="font-size: 11.5px; color: #6f6257; margin-bottom: 8px;">by {bk['author']}</p>
|
| 382 |
+
</div>
|
| 383 |
+
<div>
|
| 384 |
+
<div class="progress-bar-bg" style="margin-bottom: 6px;">
|
| 385 |
+
<div class="progress-bar-fill" style="width: {bk['percentage']}%"></div>
|
| 386 |
+
</div>
|
| 387 |
+
<div style="display: flex; justify-content: space-between; font-size: 11px; color: #6f6257;">
|
| 388 |
+
<span style="overflow: hidden; text-overflow: ellipsis; white-space: nowrap;">ποΈ {bk['voice_name']}</span>
|
| 389 |
+
</div>
|
| 390 |
+
</div>
|
| 391 |
+
</div>
|
| 392 |
+
</div>
|
| 393 |
+
"""
|
| 394 |
+
|
| 395 |
+
html += """
|
| 396 |
+
</div>
|
| 397 |
+
</div>
|
| 398 |
+
"""
|
| 399 |
+
return html
|
| 400 |
+
|
| 401 |
+
|
| 402 |
+
def generate_library_html(books_list, query="", category_filt="All", empty_mode=False):
|
| 403 |
+
if empty_mode:
|
| 404 |
+
# Compliance state: Line View empty shelf visual specifications
|
| 405 |
+
return """
|
| 406 |
+
<div style="display: flex; flex-direction: column; align-items: center; justify-content: center; padding: 64px 16px; text-align: center; max-width: 440px; margin: 0 auto;">
|
| 407 |
+
<div style="position: relative; width: 90px; height: 90px; border-radius: 50%; background: #ede8e3; border: 1px dashed #ddc1b0; display: flex; align-items: center; justify-content: center; margin-bottom: 24px;">
|
| 408 |
+
<span style="font-size: 32px;">π</span>
|
| 409 |
+
<span style="position: absolute; font-size: 20px; bottom: -4px; right: -4px;">ποΈ</span>
|
| 410 |
+
</div>
|
| 411 |
+
<h3 class="serif-header" style="font-size: 22px; margin-bottom: 8px;">Your Bookshelf is Waiting to be Voiced</h3>
|
| 412 |
+
<p style="font-size: 13.5px; color: #6f6257; line-height: 1.5; margin-bottom: 24px;">
|
| 413 |
+
Choose a comforting companion, clone their warm voice, and hear any classical or modern tale come to life immediately.
|
| 414 |
+
</p>
|
| 415 |
+
<div style="display: flex; gap: 12px; width: 100%;">
|
| 416 |
+
<span style="flex: 1; padding: 12px; background: #944a00; color: white; border-radius: 12px; font-weight: 700; font-size: 13px;">Use 'Clone Voice' tab to craft a companion</span>
|
| 417 |
+
</div>
|
| 418 |
+
</div>
|
| 419 |
+
"""
|
| 420 |
+
|
| 421 |
+
filtered = []
|
| 422 |
+
for bk in books_list:
|
| 423 |
+
match_query = (query.lower() in bk["title"].lower()) or (query.lower() in bk["author"].lower()) or (query.lower() in bk["voice_name"].lower())
|
| 424 |
+
match_cat = True
|
| 425 |
+
|
| 426 |
+
if category_filt == "In Progress":
|
| 427 |
+
match_cat = 0 < bk["percentage"] < 100
|
| 428 |
+
elif category_filt == "Completed":
|
| 429 |
+
match_cat = bk["percentage"] == 100
|
| 430 |
+
|
| 431 |
+
if match_query and match_cat:
|
| 432 |
+
filtered.append(bk)
|
| 433 |
+
|
| 434 |
+
if len(filtered) == 0:
|
| 435 |
+
return """
|
| 436 |
+
<div style="text-align: center; padding: 48px; border: 1px dashed #ebdccb; border-radius: 16px;">
|
| 437 |
+
<span style="font-size: 28px;">π</span>
|
| 438 |
+
<p style="font-size: 14px; color: #6f6257; margin-top: 8px;">No audiobooks match your search criteria.</p>
|
| 439 |
+
</div>
|
| 440 |
+
"""
|
| 441 |
+
|
| 442 |
+
html = """<div class="book-shelf-grid">"""
|
| 443 |
+
for bk in filtered:
|
| 444 |
+
html += f"""
|
| 445 |
+
<div class="shelf-card group" style="flex-direction: column; gap: 0; padding: 0; overflow: hidden;">
|
| 446 |
+
<div style="position: relative; aspect-ratio: 3/4; overflow: hidden; height: 200px;">
|
| 447 |
+
<img src="{bk['cover_url']}" style="width: 100%; height: 100%; object-fit: cover;" referrerPolicy="no-referrer" />
|
| 448 |
+
<div style="position: absolute; top: 12px; right: 12px; background: rgba(255,255,255,0.95); padding: 4px 10px; border-radius: 8px; font-weight: 700; font-size: 11px; border: 1px solid #ebdccb; color: #1c1c19;">
|
| 449 |
+
{bk['percentage']}%
|
| 450 |
+
</div>
|
| 451 |
+
<div style="position: absolute; bottom: 0; left: 0; right: 0; background: linear-gradient(0deg, rgba(0,0,0,0.85) 0%, rgba(0,0,0,0) 100%); padding: 12px; color: white;">
|
| 452 |
+
<span style="font-size: 9px; text-transform: uppercase; letter-spacing: 1.5px; font-weight: 700; color: #f5841f;">{bk['category']}</span>
|
| 453 |
+
<h5 class="serif-header" style="color: white !important; font-size: 15px; margin: 2px 0 0 0; text-overflow: ellipsis; white-space: nowrap; overflow: hidden;">{bk['title']}</h5>
|
| 454 |
+
</div>
|
| 455 |
+
</div>
|
| 456 |
+
<div style="padding: 16px; display: flex; flex-direction: column; gap: 10px;">
|
| 457 |
+
<div style="display: flex; justify-content: space-between; font-size: 11px; align-items: center;">
|
| 458 |
+
<span style="color: #6f6257;">Narrator:</span>
|
| 459 |
+
<span style="font-weight: 600; color: #944a00;">π {bk['voice_name']}</span>
|
| 460 |
+
</div>
|
| 461 |
+
<div class="progress-bar-bg">
|
| 462 |
+
<div class="progress-bar-fill" style="width: {bk['percentage']}%"></div>
|
| 463 |
+
</div>
|
| 464 |
+
<div style="display: flex; justify-content: space-between; font-size: 11px; color: #6f6257; align-items: center; padding-top: 4px;">
|
| 465 |
+
<span>{math.floor(bk['duration'] / 60)} mins length</span>
|
| 466 |
+
<span style="font-weight: 700; color: #f5841f;">Select below to Listen π§</span>
|
| 467 |
+
</div>
|
| 468 |
+
</div>
|
| 469 |
+
</div>
|
| 470 |
+
"""
|
| 471 |
+
html += "</div>"
|
| 472 |
+
return html
|
| 473 |
+
|
| 474 |
+
def render_cloned_voices_html(voices_list):
|
| 475 |
+
html = """<div class="book-shelf-grid">"""
|
| 476 |
+
for v in voices_list:
|
| 477 |
+
status_color = "#16a34a" if v["status"] == "ready" else "#d97706"
|
| 478 |
+
status_bg = "#f0fdf4" if v["status"] == "ready" else "#fef3c7"
|
| 479 |
+
|
| 480 |
+
html += f"""
|
| 481 |
+
<div class="shelf-card" style="align-items: center;">
|
| 482 |
+
<div style="width: 48px; height: 48px; border-radius: 50%; background: #ede8e3; border: 1px solid #ddc1b0; font-size: 24px; display: flex; align-items: center; justify-content: center; shrink-0: 0;">
|
| 483 |
+
{v['avatar_url']}
|
| 484 |
+
</div>
|
| 485 |
+
<div style="min-width: 0; flex: 1;">
|
| 486 |
+
<div style="display: flex; justify-content: space-between; align-items: center; gap: 6px;">
|
| 487 |
+
<h5 class="serif-header" style="font-size: 14.5px; margin: 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap;">{v['name']}</h5>
|
| 488 |
+
<span style="font-size: 9px; font-weight: 700; text-transform: uppercase; padding: 2px 6px; border-radius: 4px; color: {status_color}; background: {status_bg};">
|
| 489 |
+
{v['status']}
|
| 490 |
+
</span>
|
| 491 |
+
</div>
|
| 492 |
+
<p style="font-size: 11.5px; color: #6f6257; margin: 4px 0 0 0; line-clamp: 2; overflow: hidden; display: -webkit-box; -webkit-line-clamp: 2; -webkit-box-orient: vertical;">
|
| 493 |
+
{v['description']}
|
| 494 |
+
</p>
|
| 495 |
+
<div style="border-top: 1px solid rgba(148,74,0,0.08); padding-top: 6px; margin-top: 6px; display: flex; justify-content: space-between; font-size: 10px; color: #6f6257;">
|
| 496 |
+
<span>Gender: {v['gender']}</span>
|
| 497 |
+
<span>Created: {v['created_date']}</span>
|
| 498 |
+
</div>
|
| 499 |
+
</div>
|
| 500 |
+
</div>
|
| 501 |
+
"""
|
| 502 |
+
html += "</div>"
|
| 503 |
+
return html
|
| 504 |
+
|
| 505 |
+
|
| 506 |
+
# Gradio Application Core setup
|
| 507 |
+
with gr.Blocks(css=css_code, title="VoiceBook Gradio Hub") as demo:
|
| 508 |
+
|
| 509 |
+
# Internal variables mapping user-session states
|
| 510 |
+
books_state = gr.State(mock_books)
|
| 511 |
+
voices_state = gr.State(mock_voices)
|
| 512 |
+
|
| 513 |
+
# Simple navigation bar custom markup
|
| 514 |
+
gr.HTML("""
|
| 515 |
+
<div style="display: flex; align-items: center; justify-content: space-between; padding: 16px 0; border-bottom: 1px solid #ebdccb; margin-bottom: 24px; flex-wrap: wrap; gap: 16px;">
|
| 516 |
+
<div style="display: flex; align-items: center; gap: 12px;">
|
| 517 |
+
<div style="width: 42px; height: 42px; background-color: #944a00; border-radius: 12px; display: flex; align-items: center; justify-content: center; color: white; font-weight: bold; font-size: 20px;">
|
| 518 |
+
πΏ
|
| 519 |
+
</div>
|
| 520 |
+
<div>
|
| 521 |
+
<h1 class="serif-header" style="font-size: 22px; margin: 0; line-height: 1;">VoiceBook</h1>
|
| 522 |
+
<span style="font-size: 10px; color: #f5841f; font-weight: 700; text-transform: uppercase; letter-spacing: 1.5px;">AI Companion Companion</span>
|
| 523 |
+
</div>
|
| 524 |
+
</div>
|
| 525 |
+
<div style="display: flex; align-items: center; gap: 6px; font-size: 11.5px; color: #6f6257; font-family: monospace;">
|
| 526 |
+
<span>Hugging Face Space Node:</span>
|
| 527 |
+
<span style="color: #16a34a; font-weight: 700;">β Active</span>
|
| 528 |
+
</div>
|
| 529 |
+
</div>
|
| 530 |
+
""")
|
| 531 |
+
|
| 532 |
+
# Core Application Tabs Layout (Dashboard, Library, Clone, Player, Settings)
|
| 533 |
+
with gr.Tabs() as main_tabs:
|
| 534 |
+
|
| 535 |
+
# TAB 1: Explore Dashboard
|
| 536 |
+
with gr.TabItem("ποΈ Explore Space") as explore_tab:
|
| 537 |
+
dashboard_container = gr.HTML(value=generate_dashboard_html(mock_books))
|
| 538 |
+
|
| 539 |
+
gr.HTML("""<div style="margin-top: 24px; text-align: center;"><span style="font-size: 12px; color:#6f6257;">π§ Head over to the <b>My Bookshelf</b> or <b>Companion Player</b> tabs to stream these books with cloned pitch audio!</span></div>""")
|
| 540 |
+
|
| 541 |
+
# TAB 2: My Bookshelf
|
| 542 |
+
with gr.TabItem("π My Bookshelf") as shelf_tab:
|
| 543 |
+
with gr.Row():
|
| 544 |
+
with gr.Column(scale=3):
|
| 545 |
+
search_input = gr.Textbox(placeholder="Search by title, author, or voice...", label="Filter Library bookshelf contents", show_label=False)
|
| 546 |
+
with gr.Column(scale=1):
|
| 547 |
+
category_filter = gr.Radio(["All", "In Progress", "Completed"], value="All", label="Progress Segment", show_label=False)
|
| 548 |
+
with gr.Column(scale=1):
|
| 549 |
+
empty_state_sim = gr.Checkbox(label="Simulate Empty State", value=False)
|
| 550 |
+
|
| 551 |
+
shelf_grid = gr.HTML(value=generate_library_html(mock_books))
|
| 552 |
+
|
| 553 |
+
gr.HTML("""<div style="margin-top: 16px; border-bottom: 1px solid #ebdccb; padding-bottom: 12px;"><h4 class="serif-header" style="font-size: 16px;">Quick Mount Audiobook to Companion Player:</h4></div>""")
|
| 554 |
+
|
| 555 |
+
# Interactive launcher list so they can easily key into player
|
| 556 |
+
book_titles_list = [b["title"] for b in mock_books]
|
| 557 |
+
book_mount_selector = gr.Dropdown(choices=book_titles_list, value=book_titles_list[0], label="Choose an audiobook from your current inventory shelf", show_label=True)
|
| 558 |
+
mount_play_btn = gr.Button("π§ Mount & Launch Inside Companion Player", variant="primary")
|
| 559 |
+
|
| 560 |
+
# TAB 3: Clone Voice Studio Custom Page
|
| 561 |
+
with gr.TabItem("ποΈ Clone Voice Studio") as clone_tab:
|
| 562 |
+
with gr.Row():
|
| 563 |
+
with gr.Column(scale=1, elem_classes="card-container"):
|
| 564 |
+
gr.HTML("""
|
| 565 |
+
<h3 class="serif-header" style="font-size: 18px; margin-bottom: 4px; display: flex; align-items: center; gap: 8px;">π§ Voice Details Form</h3>
|
| 566 |
+
<p style="font-size: 12px; color: #6f6257; margin-bottom: 20px;">Provide a comforting nickname and details below to prepare vocal weights matching.</p>
|
| 567 |
+
""")
|
| 568 |
+
|
| 569 |
+
new_voice_name = gr.Textbox(placeholder="Enter a descriptive nickname (e.g., Grandma Judy, Dad)...", label="Voice Nickname")
|
| 570 |
+
new_voice_gender = gr.Radio(["Female", "Male"], value="Female", label="Voice Gender Avatar")
|
| 571 |
+
|
| 572 |
+
gr.HTML("""
|
| 573 |
+
<div style="margin: 16px 0; padding: 12px; background: white; border-radius: 12px; border: 1px solid #ebdccb; font-size: 12px;">
|
| 574 |
+
<strong style="color: #6f6257; text-transform: uppercase; font-size: 10px; display: block; margin-bottom: 4px;">Story Script to read aloud:</strong>
|
| 575 |
+
<p style="font-family: Georgia, serif; font-style: italic; color: #1c1c19; margin: 0; line-height: 1.4;">
|
| 576 |
+
"Deep within the ancient forest, the giant oak tree stood. Its roots whispered stories of forgotten eras. 'Close your eyes,' the wind hummed softly, 'for our beautiful journey begins tonight.'"
|
| 577 |
+
</p>
|
| 578 |
+
</div>
|
| 579 |
+
""")
|
| 580 |
+
|
| 581 |
+
# Record mic audio natively inside Spaces container
|
| 582 |
+
mic_recorder = gr.Audio(sources=["microphone"], type="filepath", label="Record speaking story prompt (Min 10 seconds sample)")
|
| 583 |
+
|
| 584 |
+
extract_btn = gr.Button("πͺ Analyze & Synthesize Customized Vocal Profile", variant="primary")
|
| 585 |
+
|
| 586 |
+
with gr.Column(scale=1, elem_classes="card-container"):
|
| 587 |
+
gr.HTML("""
|
| 588 |
+
<h3 class="serif-header" style="font-size: 18px; margin-bottom: 4px;">Status Pipeline Visualizer</h3>
|
| 589 |
+
<p style="font-size: 12px; color: #6f6257; margin-bottom: 16px;">Track extraction, de-noising, and pitch tuning.</p>
|
| 590 |
+
""")
|
| 591 |
+
|
| 592 |
+
# Simulated progres bars inside Gradio
|
| 593 |
+
cloning_progress_msg = gr.HTML("""
|
| 594 |
+
<div style="text-align: center; padding: 32px 0;">
|
| 595 |
+
<span style="font-size: 42px;">ποΈ</span>
|
| 596 |
+
<p style="margin-top: 12px; font-size: 14px; color: #6f6257;">Fill parameters on the left and read the story script aloud to begin extraction.</p>
|
| 597 |
+
</div>
|
| 598 |
+
""")
|
| 599 |
+
|
| 600 |
+
loading_spinner = gr.HTML(visible=False)
|
| 601 |
+
voice_cloning_success_panel = gr.HTML(visible=False)
|
| 602 |
+
|
| 603 |
+
# Mini test player for procedural sample generator preview
|
| 604 |
+
voice_sample_preview_widget = gr.Audio(value="sample_sounds/cloned_preview.wav", visible=False, label="Pre-listening synthesized vocal pitch weights match preview", interactive=False)
|
| 605 |
+
|
| 606 |
+
add_to_library_btn = gr.Button("β¨ Synced successfully! Bind Companion Voice to Inventory", visible=False)
|
| 607 |
+
|
| 608 |
+
# Bottom section detailing installed models
|
| 609 |
+
gr.HTML("""
|
| 610 |
+
<div style="margin-top: 32px; border-top: 1px solid #ebdccb; padding-top: 24px;">
|
| 611 |
+
<h4 class="serif-header" style="font-size: 18px; margin-bottom: 16px;">Currently Available Companion Voices</h4>
|
| 612 |
+
</div>
|
| 613 |
+
""")
|
| 614 |
+
cloned_voices_list_grid = gr.HTML(value=render_cloned_voices_html(mock_voices))
|
| 615 |
+
|
| 616 |
+
# TAB 4: Active Companion Player
|
| 617 |
+
with gr.TabItem("π§ Companion Player", id="player") as player_tab:
|
| 618 |
+
|
| 619 |
+
with gr.Row():
|
| 620 |
+
with gr.Column(scale=2, elem_classes="dark-card-container"):
|
| 621 |
+
gr.HTML("""
|
| 622 |
+
<div style="display: flex; justify-content: space-between; align-items: center; border-bottom: 1px dashed rgba(255,255,255,0.15); padding-bottom: 12px; margin-bottom: 16px;">
|
| 623 |
+
<span style="font-size: 10px; text-transform: uppercase; letter-spacing: 2px; color: #f5841f; font-weight: 700;">Spatial Stream Receiver Active</span>
|
| 624 |
+
<span style="background: rgba(22, 163, 74, 0.2); padding: 2px 6px; border-radius: 6px; font-size: 9px; font-family: monospace; color: #4ade80;">SYNCHRONIZED PITCH</span>
|
| 625 |
+
</div>
|
| 626 |
+
""")
|
| 627 |
+
|
| 628 |
+
player_title_info = gr.HTML("""
|
| 629 |
+
<div style="text-align: center; margin-bottom: 24px;">
|
| 630 |
+
<img src="https://images.unsplash.com/photo-1544947950-fa07a98d237f?auto=format&fit=crop&q=80&w=400" style="width: 130px; height: 180px; object-fit: cover; border-radius: 12px; box-shadow: 0 4px 16px rgba(0,0,0,0.4); margin-bottom: 16px;" />
|
| 631 |
+
<h3 style="font-family: 'Playfair Display', Georgia, serif; font-size: 22px; margin:0; color:#FAF7F2;">The Enchanted Willow</h3>
|
| 632 |
+
<p style="font-size: 12px; color: #94a3b8; font-style: italic; margin-top:2px;">by Elena S. Thorne</p>
|
| 633 |
+
<span style="display: inline-block; background: rgba(245, 132, 31, 0.15); color: #ffd3a6; font-size: 10.5px; font-weight: 700; padding: 3px 8px; border-radius: 999px; margin-top: 8px;">Active Narrator: Mom's Voice</span>
|
| 634 |
+
</div>
|
| 635 |
+
""")
|
| 636 |
+
|
| 637 |
+
player_audio_control = gr.Audio(visible=True, interactive=False, label="VoiceBook Audiobook Audio Stream Channel", value="sample_sounds/willow.wav", autoplay=False)
|
| 638 |
+
|
| 639 |
+
gr.HTML("""
|
| 640 |
+
<div style="margin-top: 16px; background: rgba(255, 255, 255, 0.05); padding: 12px; border-radius: 12px; text-align: center; font-size: 11px; color:#cbd5e1;">
|
| 641 |
+
π‘ <b>Standard Player Controls:</b> Click the standard play trigger on the audio player block above to listen to custom generated pitch chimes.
|
| 642 |
+
</div>
|
| 643 |
+
""")
|
| 644 |
+
|
| 645 |
+
# Ask / Resume / Finish controls
|
| 646 |
+
player_status_bar = gr.HTML("""
|
| 647 |
+
<div style="margin-top: 16px; display: flex; align-items: center; gap: 10px; justify-content: center;">
|
| 648 |
+
<span style="width: 8px; height: 8px; border-radius: 50%; background: #4ade80; display: inline-block;"></span>
|
| 649 |
+
<span style="font-size: 11px; color: #94a3b8; font-family: monospace;">PLAYING</span>
|
| 650 |
+
</div>
|
| 651 |
+
""")
|
| 652 |
+
|
| 653 |
+
with gr.Row():
|
| 654 |
+
ask_btn = gr.Button("β Ask a Question", variant="primary", scale=2)
|
| 655 |
+
resume_btn = gr.Button("βΆοΈ Resume Story", variant="secondary", scale=2, visible=False)
|
| 656 |
+
|
| 657 |
+
# Story-finished prompt (hidden until playback ends)
|
| 658 |
+
story_finished_panel = gr.HTML("""
|
| 659 |
+
<div style="margin-top: 12px; padding: 16px; background: rgba(245,132,31,0.1); border: 1px solid rgba(245,132,31,0.3); border-radius: 14px; text-align: center;">
|
| 660 |
+
<span style="font-size: 28px;">π</span>
|
| 661 |
+
<h4 style="font-family: 'Playfair Display', Georgia, serif; color: #ffd3a6; margin: 8px 0 4px;">Story Complete!</h4>
|
| 662 |
+
<p style="font-size: 12px; color: #94a3b8; margin-bottom: 12px;">Head to <b>My Bookshelf</b> to pick another adventure.</p>
|
| 663 |
+
</div>
|
| 664 |
+
""", visible=False)
|
| 665 |
+
|
| 666 |
+
with gr.Column(scale=3, elem_classes="card-container"):
|
| 667 |
+
gr.HTML("""
|
| 668 |
+
<h3 class="serif-header" style="font-size: 18px; margin-bottom: 4px;">Dynamic Synced Script Transcriptions</h3>
|
| 669 |
+
<p style="font-size: 12px; color: #6f6257; margin-bottom: 16px;">The synced transcript lines match progress in real-time. Use the timeline slider below to tick forward.</p>
|
| 670 |
+
""")
|
| 671 |
+
|
| 672 |
+
# Dialogue sync script box
|
| 673 |
+
synced_subtitle_ticker = gr.HTML("""
|
| 674 |
+
<div class="dialog-synced">
|
| 675 |
+
“Deep within the heart of Whispering Woods stands an ancient willow tree...”
|
| 676 |
+
</div>
|
| 677 |
+
""")
|
| 678 |
+
|
| 679 |
+
# Simulation scrub mechanism
|
| 680 |
+
timeline_slider = gr.Slider(minimum=0, maximum=5, step=1, value=0, label="Synced Script Excerpt Position (Story timeline coordinate index)")
|
| 681 |
+
|
| 682 |
+
gr.HTML("""
|
| 683 |
+
<div style="margin-top: 24px; padding-top: 16px; border-top: 1px solid #ebdccb;">
|
| 684 |
+
<h4 class="serif-header" style="font-size: 14px; margin-bottom: 12px;">Active Audiobook Metadata Details</h4>
|
| 685 |
+
</div>
|
| 686 |
+
""")
|
| 687 |
+
|
| 688 |
+
m_synopsis = gr.Textbox(interactive=False, label="Selected Story Synopsis", lines=4, value=mock_books[0]["synopsis"])
|
| 689 |
+
m_category = gr.Label(label="Story Category Shelf Mapping", value="Fantasy")
|
| 690 |
+
|
| 691 |
+
# --- Q&A Panel (hidden while playing, shown after Ask) ---
|
| 692 |
+
qa_panel = gr.HTML("""
|
| 693 |
+
<div style="margin-top: 20px; border-top: 1px solid #ebdccb; padding-top: 16px;">
|
| 694 |
+
<h4 class="serif-header" style="font-size: 14px; margin-bottom: 12px;">β Ask About the Story</h4>
|
| 695 |
+
</div>
|
| 696 |
+
""", visible=False)
|
| 697 |
+
|
| 698 |
+
with gr.Group(visible=False) as qa_input_group:
|
| 699 |
+
question_text = gr.Textbox(
|
| 700 |
+
placeholder="Type your question about the story here...",
|
| 701 |
+
label="Your Question",
|
| 702 |
+
lines=2,
|
| 703 |
+
show_label=False
|
| 704 |
+
)
|
| 705 |
+
question_audio = gr.Audio(
|
| 706 |
+
sources=["microphone"],
|
| 707 |
+
type="filepath",
|
| 708 |
+
label="Or speak your question",
|
| 709 |
+
show_label=True
|
| 710 |
+
)
|
| 711 |
+
submit_question_btn = gr.Button("π Get Answer in Cloned Voice", variant="primary")
|
| 712 |
+
|
| 713 |
+
# Answer output area
|
| 714 |
+
answer_display = gr.HTML(visible=False)
|
| 715 |
+
answer_audio = gr.Audio(
|
| 716 |
+
label="Answer in Narrator's Voice",
|
| 717 |
+
interactive=False,
|
| 718 |
+
visible=False
|
| 719 |
+
)
|
| 720 |
+
|
| 721 |
+
# TAB 5: Profile & Connection Sandbox
|
| 722 |
+
with gr.TabItem("βοΈ Profile & Sandbox Settings") as profile_tab:
|
| 723 |
+
with gr.Row():
|
| 724 |
+
with gr.Column(scale=1, elem_classes="card-container"):
|
| 725 |
+
gr.HTML("""
|
| 726 |
+
<div style="text-align: center; margin-bottom: 24px;">
|
| 727 |
+
<div style="width: 72px; height: 72px; border-radius: 50%; background: #944a00; color: white; font-size: 36px; display: flex; align-items: center; justify-content: center; margin: 0 auto 12px auto;">
|
| 728 |
+
π©βπ»
|
| 729 |
+
</div>
|
| 730 |
+
<h3 class="serif-header" style="font-size: 18px; margin: 0;">Sarah Jenkins</h3>
|
| 731 |
+
<p style="font-size: 11.5px; color: #6f6257; margin-top: 2px;">Premium Member since Summer 2026</p>
|
| 732 |
+
</div>
|
| 733 |
+
""")
|
| 734 |
+
|
| 735 |
+
gr.HTML("""
|
| 736 |
+
<div style="display: grid; grid-template-columns: 1fr 1fr 1fr; gap: 8px; text-align: center;">
|
| 737 |
+
<div style="background: white; border: 1px solid #ebdccb; border-radius: 12px; padding: 10px;">
|
| 738 |
+
<span style="font-size: 18px; display: block;">π</span>
|
| 739 |
+
<strong style="font-size: 15px; display: block; color: #1c1c19;">5</strong>
|
| 740 |
+
<span style="font-size: 9px; text-transform: uppercase; color: #6f6257;">Bookshelf</span>
|
| 741 |
+
</div>
|
| 742 |
+
<div style="background: white; border: 1px solid #ebdccb; border-radius: 12px; padding: 10px;">
|
| 743 |
+
<span style="font-size: 18px; display: block;">β±οΈ</span>
|
| 744 |
+
<strong style="font-size: 15px; display: block; color: #1c1c19;">14.6h</strong>
|
| 745 |
+
<span style="font-size: 9px; text-transform: uppercase; color: #6f6257;">Listen Time</span>
|
| 746 |
+
</div>
|
| 747 |
+
<div style="background: white; border: 1px solid #ebdccb; border-radius: 12px; padding: 10px;">
|
| 748 |
+
<span style="font-size: 18px; display: block;">ποΈ</span>
|
| 749 |
+
<strong style="font-size: 15px; display: block; color: #1c1c19;">3</strong>
|
| 750 |
+
<span style="font-size: 9px; text-transform: uppercase; color: #6f6257;">Cloned Cast</span>
|
| 751 |
+
</div>
|
| 752 |
+
</div>
|
| 753 |
+
""")
|
| 754 |
+
|
| 755 |
+
gr.HTML("""
|
| 756 |
+
<div style="margin-top: 24px;">
|
| 757 |
+
<h4 class="serif-header" style="font-size: 14px; margin-bottom: 8px;">Acoustic Signal Processing Config</h4>
|
| 758 |
+
</div>
|
| 759 |
+
""")
|
| 760 |
+
gr.Checkbox(label="Enable Client Denoising Filter", value=True)
|
| 761 |
+
gr.Checkbox(label="Enable Brownian Dynamic Sleep Wave", value=False)
|
| 762 |
+
gr.Checkbox(label="Automatic Chapter Transitions", value=True)
|
| 763 |
+
|
| 764 |
+
with gr.Column(scale=1, elem_classes="card-container"):
|
| 765 |
+
gr.HTML("""
|
| 766 |
+
<h3 class="serif-header" style="font-size: 18px; margin-bottom: 4px;">Boundary Connection Error Diagnostics</h3>
|
| 767 |
+
<p style="font-size: 12px; color: #6f6257; margin-bottom: 16px;">Test robust offline error conditions gracefully.</p>
|
| 768 |
+
""")
|
| 769 |
+
|
| 770 |
+
offline_toggle = gr.Checkbox(label="Simulate Sandbox Connection Outage", value=False)
|
| 771 |
+
|
| 772 |
+
gr.HTML("""
|
| 773 |
+
<div style="margin-top: 16px; padding: 12px; background: white; border-radius: 12px; border: 1px solid #ebdccb;">
|
| 774 |
+
<span style="font-size:10px; font-weight:700; text-transform:uppercase; color:#6f6257; display:block; margin-bottom:6px;">Sandbox local metric stores:</span>
|
| 775 |
+
<div style="font-size: 11.5px; color: #1c1c19;">
|
| 776 |
+
πΎ Voice Weights Space Cache: 24.8 MB / 512 MB<br>
|
| 777 |
+
π¦ Cached offline items: 182 MB (12%)
|
| 778 |
+
</div>
|
| 779 |
+
</div>
|
| 780 |
+
""")
|
| 781 |
+
|
| 782 |
+
error_sim_view = gr.HTML("""
|
| 783 |
+
<div style="margin-top: 16px; padding: 16px; background: #fef2f2; border: 1px solid #fecaca; border-radius: 12px; display: none;" id="error-sim-panel">
|
| 784 |
+
<span style="font-size: 20px; display: block; margin-bottom: 4px;">β οΈ Connection Interruption Layout</span>
|
| 785 |
+
<strong style="color: #991b1b; font-size:14px; display:block;">Can't reach the story right now</strong>
|
| 786 |
+
<p style="color: #7f1d1d; font-size: 11.5px; margin-top:4px; line-height:1.4;">The signal to Whispering Woods took a brief tumble. Verify your client network link or refresh the bookshelf items.</p>
|
| 787 |
+
</div>
|
| 788 |
+
""", elem_id="error-sim-block")
|
| 789 |
+
|
| 790 |
+
# Offline outage container modal overlay representation
|
| 791 |
+
system_error_modal = gr.HTML("""
|
| 792 |
+
<div style="display: none; background-color: #FAF7F2; padding: 48px 16px; text-align: center; border-radius: 20px; border: 2px solid #fecaca; max-width: 460px; margin: 40px auto;" id="global-sandbox-error">
|
| 793 |
+
<div style="width: 72px; height: 72px; border-radius: 16px; background: #fff5f5; border: 1px solid #fecaca; display: flex; align-items: center; justify-content: center; margin: 0 auto 20px auto; font-size: 36px;">
|
| 794 |
+
β οΈ
|
| 795 |
+
</div>
|
| 796 |
+
<span style="font-size: 10px; font-weight: 700; color: #dc2626; text-transform: uppercase; letter-spacing: 2px;">Offline Connection Interruption</span>
|
| 797 |
+
<h2 class="serif-header" style="font-size: 24px; margin: 8px 0;">Can't reach the story right now</h2>
|
| 798 |
+
<p style="font-size: 13.5px; color: #6f6257; line-height: 1.5; margin-bottom: 24px;">The signal to Whispering Woods took a brief tumble. Verify your client network link or try refreshing.</p>
|
| 799 |
+
<div style="background: #ede8e3; padding: 10px; border-radius: 10px; font-size: 11px; color:#6f6257; font-family: monospace;">
|
| 800 |
+
Boundary Sandbox Mode Enabled. Uncheck settings switch to resume.
|
| 801 |
+
</div>
|
| 802 |
+
</div>
|
| 803 |
+
""", visible=False)
|
| 804 |
+
|
| 805 |
+
# REACTIVE LOGIC TRIGGERS & Callbacks
|
| 806 |
+
|
| 807 |
+
# 1. Search, filter & simulate empty state on shelf
|
| 808 |
+
def filter_books_shelf(query, category, empty_mode_active, current_books):
|
| 809 |
+
if empty_mode_active:
|
| 810 |
+
# Render special formatted zero-state visual
|
| 811 |
+
return generate_library_html(current_books, empty_mode=True)
|
| 812 |
+
return generate_library_html(current_books, query=query, category_filt=category)
|
| 813 |
+
|
| 814 |
+
search_input.change(filter_books_shelf, inputs=[search_input, category_filter, empty_state_sim, books_state], outputs=shelf_grid)
|
| 815 |
+
category_filter.change(filter_books_shelf, inputs=[search_input, category_filter, empty_state_sim, books_state], outputs=shelf_grid)
|
| 816 |
+
empty_state_sim.change(filter_books_shelf, inputs=[search_input, category_filter, empty_state_sim, books_state], outputs=shelf_grid)
|
| 817 |
+
|
| 818 |
+
# 2. Extract vocal weights trigger (wizard pipeline progress bar matching)
|
| 819 |
+
def animate_cloning_pipeline(v_name, v_gender, recorder_data, progress=gr.Progress()):
|
| 820 |
+
if not v_name.strip():
|
| 821 |
+
return (
|
| 822 |
+
"""<div style="padding: 12px; background: #fef2f2; border: 1px solid #fecaca; color: #991b1b; border-radius: 12px; font-size: 12px; font-weight: 600;">
|
| 823 |
+
β οΈ Please provide a comforting nickname for your cloning candidate.
|
| 824 |
+
</div>""",
|
| 825 |
+
gr.HTML(visible=False),
|
| 826 |
+
gr.HTML(visible=False),
|
| 827 |
+
gr.Audio(visible=False),
|
| 828 |
+
gr.Button(visible=False)
|
| 829 |
+
)
|
| 830 |
+
if recorder_data is None:
|
| 831 |
+
return (
|
| 832 |
+
"""<div style="padding: 12px; background: #fef2f2; border: 1px solid #fecaca; color: #991b1b; border-radius: 12px; font-size: 12px; font-weight: 600;">
|
| 833 |
+
β οΏ½οΏ½ Microphone recording sample missing. Speak some script lines!
|
| 834 |
+
</div>""",
|
| 835 |
+
gr.HTML(visible=False),
|
| 836 |
+
gr.HTML(visible=False),
|
| 837 |
+
gr.Audio(visible=False),
|
| 838 |
+
gr.Button(visible=False)
|
| 839 |
+
)
|
| 840 |
+
|
| 841 |
+
progress(0, desc="Extracting speech samples patterns...")
|
| 842 |
+
time.sleep(1.0)
|
| 843 |
+
progress(0.3, desc="Filtering ambient air-con noise frequencies...")
|
| 844 |
+
time.sleep(1.2)
|
| 845 |
+
progress(0.65, desc="Mapping vocal vocal tracts formant spaces & frequencies...")
|
| 846 |
+
time.sleep(1.5)
|
| 847 |
+
progress(0.9, desc="Validating speech prosody & standard conversational matching...")
|
| 848 |
+
time.sleep(1.0)
|
| 849 |
+
|
| 850 |
+
preview_text = f"“Hello! I have created my brand new clone under '{v_name}'. Ready to narrate any classical story inside your library shelves.”"
|
| 851 |
+
|
| 852 |
+
avatar = "π©" if v_gender == "Female" else "π¨"
|
| 853 |
+
|
| 854 |
+
cloned_card_html = f"""
|
| 855 |
+
<div style="background: white; border: 1px solid #ebdccb; padding: 24px; border-radius: 16px; margin: 16px 0;">
|
| 856 |
+
<div style="display: flex; gap: 16px; align-items: center; margin-bottom: 12px;">
|
| 857 |
+
<div style="width: 48px; height: 48px; border-radius: 50%; background: #944a00; color: white; display: flex; align-items: center; justify-content: center; font-size: 24px;">
|
| 858 |
+
{avatar}
|
| 859 |
+
</div>
|
| 860 |
+
<div>
|
| 861 |
+
<h4 class="serif-header" style="font-size: 16px; margin: 0;">{v_name} <span style="font-size: 9px; background: #f0fdf4; color: #16a34a; padding: 2px 6px; border-radius: 4px; uppercase: text-transform; font-weight:700;">Cloned successfully</span></h4>
|
| 862 |
+
<p style="font-size: 11px; color:#6f6257; margin-top:2px;">Synthesized today β’ Calibrated for Classical Audiobooks</p>
|
| 863 |
+
</div>
|
| 864 |
+
</div>
|
| 865 |
+
<div style="background:#FAF7F2; border-left: 3px solid #f5841f; padding: 10px; font-family: Georgia, serif; font-style: italic; font-size: 12px; color: #1c1c19;">
|
| 866 |
+
{preview_text}
|
| 867 |
+
</div>
|
| 868 |
+
</div>
|
| 869 |
+
"""
|
| 870 |
+
|
| 871 |
+
return (
|
| 872 |
+
gr.HTML(visible=False),
|
| 873 |
+
gr.HTML(visible=False),
|
| 874 |
+
gr.HTML(value=cloned_card_html, visible=True),
|
| 875 |
+
gr.Audio(visible=True),
|
| 876 |
+
gr.Button(visible=True)
|
| 877 |
+
)
|
| 878 |
+
|
| 879 |
+
extract_btn.click(
|
| 880 |
+
animate_cloning_pipeline,
|
| 881 |
+
inputs=[new_voice_name, new_voice_gender, mic_recorder],
|
| 882 |
+
outputs=[cloning_progress_msg, loading_spinner, voice_cloning_success_panel, voice_sample_preview_widget, add_to_library_btn]
|
| 883 |
+
)
|
| 884 |
+
|
| 885 |
+
# 3. Add successfully cloned voice to inventory
|
| 886 |
+
def save_new_cloned_voice(v_name, v_gender, list_voices):
|
| 887 |
+
avatar = "π©" if v_gender == "Female" else "π¨"
|
| 888 |
+
new_voice_item = {
|
| 889 |
+
"id": f"v-{time.time()}",
|
| 890 |
+
"name": v_name,
|
| 891 |
+
"avatar_url": avatar,
|
| 892 |
+
"description": f"Custom cloned {v_gender.lower()} companion voice. Highly optimized.",
|
| 893 |
+
"gender": v_gender,
|
| 894 |
+
"created_date": "2026-06-06",
|
| 895 |
+
"status": "ready"
|
| 896 |
+
}
|
| 897 |
+
updated = [new_voice_item] + list_voices
|
| 898 |
+
|
| 899 |
+
# We also need to map this voice option dynamically into the books dropdown so they can narrate books with it!
|
| 900 |
+
return (
|
| 901 |
+
updated,
|
| 902 |
+
render_cloned_voices_html(updated),
|
| 903 |
+
gr.HTML(visible=True, value="""<div style="text-align: center; padding: 32px 0;"><span style="font-size:42px;">ποΈ</span><p style="margin-top:12px; font-size:14px; color:#6f6257;">Vocal profile saved successfully! Synthesize another one on the left if you'd like.</p></div>"""),
|
| 904 |
+
gr.HTML(visible=False),
|
| 905 |
+
gr.HTML(visible=False),
|
| 906 |
+
gr.Audio(visible=False),
|
| 907 |
+
gr.Button(visible=False),
|
| 908 |
+
gr.Textbox(value="")
|
| 909 |
+
)
|
| 910 |
+
|
| 911 |
+
add_to_library_btn.click(
|
| 912 |
+
save_new_cloned_voice,
|
| 913 |
+
inputs=[new_voice_name, new_voice_gender, voices_state],
|
| 914 |
+
outputs=[voices_state, cloned_voices_list_grid, cloning_progress_msg, loading_spinner, voice_cloning_success_panel, voice_sample_preview_widget, add_to_library_btn, new_voice_name]
|
| 915 |
+
)
|
| 916 |
+
|
| 917 |
+
# 4. Mount audiobook metadata details into player layout
|
| 918 |
+
def handle_book_mounting(title_chosen, current_inventory):
|
| 919 |
+
# find book
|
| 920 |
+
selected = next((b for b in current_inventory if b["title"] == title_chosen), current_inventory[0])
|
| 921 |
+
|
| 922 |
+
player_html_markup = f"""
|
| 923 |
+
<div style="text-align: center; margin-bottom: 24px;">
|
| 924 |
+
<img src="{selected['cover_url']}" style="width: 130px; height: 180px; object-fit: cover; border-radius: 12px; box-shadow: 0 4px 16px rgba(0,0,0,0.4); margin-bottom: 16px;" />
|
| 925 |
+
<h3 style="font-family: 'Playfair Display', Georgia, serif; font-size: 22px; margin:0; color:#FAF7F2;">{selected['title']}</h3>
|
| 926 |
+
<p style="font-size: 12px; color: #94a3b8; font-style: italic; margin-top:2px;">by {selected['author']}</p>
|
| 927 |
+
<span style="display: inline-block; background: rgba(245, 132, 31, 0.15); color: #ffd3a6; font-size: 10.5px; font-weight: 700; padding: 3px 8px; border-radius: 999px; margin-top: 8px;">Active Narrator: {selected['voice_name']}</span>
|
| 928 |
+
</div>
|
| 929 |
+
"""
|
| 930 |
+
|
| 931 |
+
# update player tabs active
|
| 932 |
+
return (
|
| 933 |
+
player_html_markup,
|
| 934 |
+
selected["audio_path"],
|
| 935 |
+
selected["synopsis"],
|
| 936 |
+
selected["category"],
|
| 937 |
+
gr.Tabs(selected="player") # Select player tab
|
| 938 |
+
)
|
| 939 |
+
|
| 940 |
+
mount_play_btn.click(
|
| 941 |
+
handle_book_mounting,
|
| 942 |
+
inputs=[book_mount_selector, books_state],
|
| 943 |
+
outputs=[player_title_info, player_audio_control, m_synopsis, m_category, main_tabs]
|
| 944 |
+
)
|
| 945 |
+
|
| 946 |
+
# 5. Playback timeline slider sync transcript subtitle lyrics
|
| 947 |
+
def update_synced_subtitles(percentage_index_choice):
|
| 948 |
+
idx = int(percentage_index_choice)
|
| 949 |
+
txt_excerpt = narrative_subtitles[idx % len(narrative_subtitles)]
|
| 950 |
+
html_markup = f"""
|
| 951 |
+
<div class="dialog-synced">
|
| 952 |
+
“{txt_excerpt}”
|
| 953 |
+
</div>
|
| 954 |
+
"""
|
| 955 |
+
return html_markup
|
| 956 |
+
|
| 957 |
+
timeline_slider.change(
|
| 958 |
+
update_synced_subtitles,
|
| 959 |
+
inputs=[timeline_slider],
|
| 960 |
+
outputs=[synced_subtitle_ticker]
|
| 961 |
+
)
|
| 962 |
+
|
| 963 |
+
# 6. Ask button β pauses story, reveals Q&A panel
|
| 964 |
+
def enter_asking_state():
|
| 965 |
+
status_html = """
|
| 966 |
+
<div style="margin-top: 16px; display: flex; align-items: center; gap: 10px; justify-content: center;">
|
| 967 |
+
<span style="width: 8px; height: 8px; border-radius: 50%; background: #f59e0b; display: inline-block; animation: pulse 1s infinite;"></span>
|
| 968 |
+
<span style="font-size: 11px; color: #f59e0b; font-family: monospace;">PAUSED β ASKING</span>
|
| 969 |
+
</div>
|
| 970 |
+
"""
|
| 971 |
+
return (
|
| 972 |
+
status_html, # player_status_bar
|
| 973 |
+
gr.Button(visible=False), # ask_btn hide
|
| 974 |
+
gr.Button(visible=True), # resume_btn show
|
| 975 |
+
gr.HTML(visible=True), # qa_panel show
|
| 976 |
+
gr.Group(visible=True), # qa_input_group show
|
| 977 |
+
gr.HTML(visible=False), # answer_display reset
|
| 978 |
+
gr.Audio(visible=False), # answer_audio reset
|
| 979 |
+
)
|
| 980 |
+
|
| 981 |
+
ask_btn.click(
|
| 982 |
+
enter_asking_state,
|
| 983 |
+
inputs=[],
|
| 984 |
+
outputs=[player_status_bar, ask_btn, resume_btn, qa_panel, qa_input_group, answer_display, answer_audio]
|
| 985 |
+
)
|
| 986 |
+
|
| 987 |
+
# 7. Submit question β generate mock answer (placeholder for real Qwen call)
|
| 988 |
+
def handle_question_submit(question_txt, question_audio_path):
|
| 989 |
+
if not question_txt.strip() and question_audio_path is None:
|
| 990 |
+
answer_html = """
|
| 991 |
+
<div style="padding: 10px; background: #fef2f2; border: 1px solid #fecaca; border-radius: 10px; font-size: 12px; color: #991b1b;">
|
| 992 |
+
β οΈ Please type a question or record one with the microphone.
|
| 993 |
+
</div>
|
| 994 |
+
"""
|
| 995 |
+
return answer_html, gr.Audio(visible=False)
|
| 996 |
+
|
| 997 |
+
q_text = question_txt.strip() if question_txt.strip() else "(spoken question)"
|
| 998 |
+
|
| 999 |
+
# Placeholder: real Qwen2.5-3B-Instruct call goes here
|
| 1000 |
+
mock_answer = f"That's a great question! The story tells us something very special about that. Keep listening to discover the full answer as the adventure unfolds."
|
| 1001 |
+
|
| 1002 |
+
answer_html = f"""
|
| 1003 |
+
<div style="margin-top: 12px; padding: 16px; background: #f0fdf4; border: 1px solid #bbf7d0; border-radius: 14px;">
|
| 1004 |
+
<div style="font-size: 10px; font-weight: 700; text-transform: uppercase; color: #16a34a; margin-bottom: 6px;">Answer in Narrator's Voice</div>
|
| 1005 |
+
<div style="font-family: 'Playfair Display', Georgia, serif; font-style: italic; color: #1c1c19; font-size: 13px; line-height: 1.6;">
|
| 1006 |
+
“{mock_answer}”
|
| 1007 |
+
</div>
|
| 1008 |
+
<div style="margin-top: 8px; font-size: 10px; color: #6f6257;">
|
| 1009 |
+
Q: <em>{q_text}</em>
|
| 1010 |
+
</div>
|
| 1011 |
+
</div>
|
| 1012 |
+
"""
|
| 1013 |
+
# answer_audio will be wired to real TTS output; for now surfaces the preview chime
|
| 1014 |
+
return answer_html, gr.Audio(value="sample_sounds/cloned_preview.wav", visible=True)
|
| 1015 |
+
|
| 1016 |
+
submit_question_btn.click(
|
| 1017 |
+
handle_question_submit,
|
| 1018 |
+
inputs=[question_text, question_audio],
|
| 1019 |
+
outputs=[answer_display, answer_audio]
|
| 1020 |
+
)
|
| 1021 |
+
|
| 1022 |
+
# 8. Resume button β restores playing state, hides Q&A panel
|
| 1023 |
+
def enter_resume_state():
|
| 1024 |
+
status_html = """
|
| 1025 |
+
<div style="margin-top: 16px; display: flex; align-items: center; gap: 10px; justify-content: center;">
|
| 1026 |
+
<span style="width: 8px; height: 8px; border-radius: 50%; background: #4ade80; display: inline-block;"></span>
|
| 1027 |
+
<span style="font-size: 11px; color: #94a3b8; font-family: monospace;">PLAYING</span>
|
| 1028 |
+
</div>
|
| 1029 |
+
"""
|
| 1030 |
+
return (
|
| 1031 |
+
status_html, # player_status_bar
|
| 1032 |
+
gr.Button(visible=True), # ask_btn show
|
| 1033 |
+
gr.Button(visible=False), # resume_btn hide
|
| 1034 |
+
gr.HTML(visible=False), # qa_panel hide
|
| 1035 |
+
gr.Group(visible=False), # qa_input_group hide
|
| 1036 |
+
gr.HTML(visible=False), # answer_display hide
|
| 1037 |
+
gr.Audio(visible=False), # answer_audio hide
|
| 1038 |
+
gr.Textbox(value=""), # question_text clear
|
| 1039 |
+
)
|
| 1040 |
+
|
| 1041 |
+
resume_btn.click(
|
| 1042 |
+
enter_resume_state,
|
| 1043 |
+
inputs=[],
|
| 1044 |
+
outputs=[player_status_bar, ask_btn, resume_btn, qa_panel, qa_input_group, answer_display, answer_audio, question_text]
|
| 1045 |
+
)
|
| 1046 |
+
|
| 1047 |
+
# 9. Story-finished trigger β slider reaches end (index 5 = last subtitle)
|
| 1048 |
+
def check_story_finished(slider_val):
|
| 1049 |
+
if int(slider_val) >= 5:
|
| 1050 |
+
return gr.HTML(visible=True)
|
| 1051 |
+
return gr.HTML(visible=False)
|
| 1052 |
+
|
| 1053 |
+
timeline_slider.change(
|
| 1054 |
+
check_story_finished,
|
| 1055 |
+
inputs=[timeline_slider],
|
| 1056 |
+
outputs=[story_finished_panel]
|
| 1057 |
+
)
|
| 1058 |
+
|
| 1059 |
+
# 10. Global Sandbox connection error boundary state toggle
|
| 1060 |
+
def toggle_outage_sandbox(checked):
|
| 1061 |
+
if checked:
|
| 1062 |
+
return (
|
| 1063 |
+
gr.HTML(visible=True), # Show error simulated overlay
|
| 1064 |
+
gr.Tabs(visible=False) # Hide whole actual menu tab structure
|
| 1065 |
+
)
|
| 1066 |
+
else:
|
| 1067 |
+
return (
|
| 1068 |
+
gr.HTML(visible=False), # Hide
|
| 1069 |
+
gr.Tabs(visible=True) # Full tabs display
|
| 1070 |
+
)
|
| 1071 |
+
|
| 1072 |
+
offline_toggle.change(
|
| 1073 |
+
toggle_outage_sandbox,
|
| 1074 |
+
inputs=[offline_toggle],
|
| 1075 |
+
outputs=[system_error_modal, main_tabs]
|
| 1076 |
+
)
|
| 1077 |
+
|
| 1078 |
+
# Launch parameters configured for standard port and auto binding
|
| 1079 |
+
if __name__ == "__main__":
|
| 1080 |
+
demo.queue().launch()
|