Describe the Bug
When using the local TTS backend with the kokoro-en-v0_19 model, ffplay exits with code 1 during audio playback, causing the TTS functionality to fail. The issue is resolved by forcing the use
of paplay instead. Note that this issue is specific to TTS (Text-to-Speech) and does not affect STT (Speech-to-Text).
Environment
- OS: Ubuntu 22.04
- Pi version: 0.71.1
- pi-listen version: 15.3
- Terminal: default ubuntu terminal
To Reproduce
Steps to reproduce the behavior:
- Install the pi-listen extension.
- Configure the local TTS backend with the kokoro-en-v0_19 model.
- Attempt to synthesize and play audio using the speak function.
- Observe that ffplay exits with code 1, and audio playback fails.
Expected Behavior
Audio should play successfully using ffplay with the kokoro-en-v0_19 model.
Actual Behavior
ffplay exits with code 1, and the following error is displayed:
Error: Streaming player ffplay exited with code 1
Diagnostic Output
-> I'm using only tts+kokoro, stt works fine, I tried to debug the extension with pi
/voice info
Voice diagnostics:
Backend: local
Audio capture:
tool: sox (rec)
local model: parakeet-v3
local endpoint: http://localhost:8080
Config:
language: en
onboarding: complete
hold threshold: 700ms
toggle shortcut: ctrl+shift+v
kitty protocol: not detected
state: idle
mic capture: FAILED — sox ran but no audio captured
sherpa-onnx: OK (in-process mode)
-> mic capture works.
Additional Context
- Node.js Version: v24.15.0
- FFmpeg Version: 4.4.2-0ubuntu0.22.04.1
- Workaround: Forcing the use of paplay resolves the issue.
- Model Used: kokoro-en-v0_19 (TTS only, not STT).
- Note: The issue is specific to TTS and does not affect STT functionality.
Describe the Bug
When using the local TTS backend with the kokoro-en-v0_19 model, ffplay exits with code 1 during audio playback, causing the TTS functionality to fail. The issue is resolved by forcing the use
of paplay instead. Note that this issue is specific to TTS (Text-to-Speech) and does not affect STT (Speech-to-Text).
Environment
To Reproduce
Steps to reproduce the behavior:
Expected Behavior
Audio should play successfully using ffplay with the kokoro-en-v0_19 model.
Actual Behavior
ffplay exits with code 1, and the following error is displayed:
Diagnostic Output
-> I'm using only tts+kokoro, stt works fine, I tried to debug the extension with pi
/voice info
Voice diagnostics:
Backend: local
Audio capture:
tool: sox (rec)
local model: parakeet-v3
local endpoint: http://localhost:8080
Config:
language: en
onboarding: complete
hold threshold: 700ms
toggle shortcut: ctrl+shift+v
kitty protocol: not detected
state: idle
mic capture: FAILED — sox ran but no audio captured
sherpa-onnx: OK (in-process mode)
-> mic capture works.
Additional Context