piper_engine.py

# piper_engine.py
import io
import numpy as np
import soundfile as sf
from piper.voice import PiperVoice

class PiperEngine:

    def __init__(self, model_path):
        # Load the voice model
        self.voice = PiperVoice.load(model_path)

    def synthesize(
        self,
        text,
        speed=1.0,   # Default speed multiplier
        pitch=1.0,   # Default pitch multiplier
        sample_rate=22050,
    ):

        # -----------------------------
        # Important: call synthesize EXACTLY as Piper expects
        # -----------------------------
        # Positional arguments only: text, length_scale, pitch_scale
        # Do not use keywords like length_scale= or pitch_scale=
        audio = self.voice.synthesize(text, speed, pitch)

        # -----------------------------
        # Convert to NumPy array and ensure correct shape
        # -----------------------------
        audio = np.array(audio, dtype=np.float32)
        if audio.ndim == 1:
            audio = audio[:, np.newaxis]

        # -----------------------------
        # Write to WAV buffer
        # -----------------------------
        buffer = io.BytesIO()
        sf.write(buffer, audio, sample_rate, format="WAV")
        buffer.seek(0)

        return buffer