examples/pip_example.py

"""
Chiluka TTS - pip install Example

After installing via pip, model weights auto-download from HuggingFace
on first use and are cached locally.

Install:
    pip install chiluka
    sudo apt-get install espeak-ng

Usage:
    python pip_example.py --reference path/to/reference.wav
    python pip_example.py --reference ref.wav --model telugu --language te
"""

import argparse


def main():
    parser = argparse.ArgumentParser(description="Chiluka TTS - pip Example")
    parser.add_argument("--reference", type=str, required=True, help="Path to reference audio file")
    parser.add_argument("--model", type=str, default="hindi_english", choices=["hindi_english", "telugu"],
                        help="Model variant (default: hindi_english)")
    parser.add_argument("--text", type=str, default=None, help="Text to synthesize")
    parser.add_argument("--language", type=str, default=None, help="Language code (en-us, hi, te)")
    parser.add_argument("--output", type=str, default="output_pip.wav", help="Output wav file path")
    args = parser.parse_args()

    # Import after argparse so --help is fast
    from chiluka import Chiluka, list_models

    # Set defaults
    if args.text is None:
        texts = {
            "hindi_english": "Hello, I am Chiluka, a text to speech system.",
            "telugu": "నమస్కారం, నేను చిలుక మాట్లాడుతున్నాను",
        }
        args.text = texts[args.model]

    if args.language is None:
        langs = {"hindi_english": "en-us", "telugu": "te"}
        args.language = langs[args.model]

    # List models
    print("Available models:")
    for name, info in list_models().items():
        print(f"  {name}: {info['description']}")
    print()

    # Load model (auto-downloads weights on first run)
    print(f"Loading '{args.model}' model...")
    tts = Chiluka.from_pretrained(model=args.model)

    # Synthesize speech
    print(f"Text: '{args.text}'")
    print(f"Language: {args.language}")
    print(f"Reference: {args.reference}")
    print()

    wav = tts.synthesize(
        text=args.text,
        reference_audio=args.reference,
        language=args.language,
        alpha=0.3,
        beta=0.7,
        diffusion_steps=5,
        embedding_scale=1.0,
    )

    # Save output
    tts.save_wav(wav, args.output)
    print(f"Duration: {len(wav) / 24000:.2f} seconds")

    # --- Bonus: synthesize in another language with same model ---
    if args.model == "hindi_english":
        print("\n--- Bonus: Hindi synthesis with same model ---")
        hindi_wav = tts.synthesize(
            text="नमस्ते, मैं चिलुका बोल रहा हूं",
            reference_audio=args.reference,
            language="hi",
        )
        hindi_output = args.output.replace(".wav", "_hindi.wav")
        tts.save_wav(hindi_wav, hindi_output)
        print(f"Duration: {len(hindi_wav) / 24000:.2f} seconds")


if __name__ == "__main__":
    main()