Synthesize audio from text using piper-tts from a onnx model

gistfile1.txt

import argparse
import wave
from pathlib import Path
from piper.voice import PiperVoice


def synthesize_speech(
    model_path: Path, model_json_path: Path, text: str, output_path: Path
):
    # Load the voice model
    voice = PiperVoice.load(model_path, model_json_path)

    fh = None
    # Synthesize the text
    for chunk in voice.synthesize(str(text)):
        if fh is None:
            fh = wave.open(str(output_path), "wb")
            fh.setframerate(chunk.sample_rate)
            fh.setsampwidth(chunk.sample_width)
            fh.setnchannels(chunk.sample_channels)
        fh.writeframes(chunk.audio_int16_bytes)

    if fh is not None:
        fh.close()
        print(f"Audio saved to {output_path}")


def main():
    # Create argument parser
    parser = argparse.ArgumentParser(
        description="Synthesize speech from text using piper-tts"
    )

    # Add arguments
    parser.add_argument(
        "--model",
        type=Path,
        default="./codsworth.onnx",
        help="Path to the ONNX voice model file",
    )

    parser.add_argument(
        "--config",
        type=Path,
        default="./codsworth.onnx.json",
        help="Path to the model's JSON configuration file",
    )

    parser.add_argument(
        "--text", type=Path, required=True, help="Text to synthesize into speech"
    )

    parser.add_argument(
        "--output", type=Path, default=Path("output.wav"), help="Output WAV file path"
    )

    # Parse arguments
    args = parser.parse_args()

    model_path = Path(args.model)
    config_path = Path(args.config)

    # Validate model files exist
    if not model_path.exists() and model_path.is_file():
        print(f"Error: Model file '{args.model}' not found")
        return

    if not config_path.exists() and config_path.is_file():
        print(f"Error: Config file '{args.config}' not found")
        return

    output_path = Path(args.output)
    output_path.parent.mkdir(parents=True, exist_ok=True)

    # Synthesize speech
    synthesize_speech(model_path, config_path, args.text, output_path)


if __name__ == "__main__":
    main()