Python API - working streaming example?

[pilgrimwebshop]

I've looked at the documentation and found the following code:

for chunk in voice.synthesize("..."):
    set_audio_format(chunk.sample_rate, chunk.sample_width, chunk.sample_channels)
    write_raw_data(chunk.audio_int16_bytes)

set_audio_format() and write_raw_data() both are undefined. Is there something I need to import, or can I get some help defining these functions?

Thank you!

[sueun-dev]

hi, those two names are placeholders for whatever audio sink you're using, not Piper functions you import.

I tried the streaming path with the repo's test voice, and writing chunks like this produced the same WAV bytes as voice.synthesize_wav(...):

import wave
from piper import PiperVoice

voice = PiperVoice.load("/path/to/voice.onnx")

with wave.open("out.wav", "wb") as wav:
    first = True
    for chunk in voice.synthesize("text to speak"):
        if first:
            wav.setframerate(chunk.sample_rate)
            wav.setsampwidth(chunk.sample_width)
            wav.setnchannels(chunk.sample_channels)
            first = False

        wav.writeframes(chunk.audio_int16_bytes)

So set_audio_format(...) means configure your sink once from the first chunk, and write_raw_data(...) means push chunk.audio_int16_bytes into that sink.

If you want actual speaker playback instead of a WAV file, #85 has a PyAudio version that does the same thing: #85 (comment)