mlx-audio · Issues· 106 open

streaming input architecture

Whisper checkpoints without a HF processor (e.g. mlx-community/whisper-large-v3-turbo-8bit, converted with mlx-audio-plus) fail late with a confusing "Processor not found" error instead of a clear load-time message

nemotron_asr: expose live-input sessions through the existing realtime STT contract

Feature: Dynamic UI forms from model metadata

Qwen3-ASR mishandles invalid, short, and no-speech audio inputs

Qwen3-ASR streaming timestamps depend on max_tokens instead of audio time

ICL voice cloning randomly degenerates (no EOS, 24s garbage) — root cause: ref_text not covering full ref_audio; confirmed not a port bug via official PyTorch A/B

Whisper non_speech_tokens crashes with IndexError: encode(" -")[0] on empty result (0.4.4 STT)

Voxtral TTS: decorative single quotes can add hallucinated speech

Model request + implementation plan: dots.tts (rednote-hilab) zero-shot voice cloning

STT load_model silently returns a randomly initialized model when checkpoint keys don't match (strict=False default)

Streaming `/v1/audio/speech` emits one complete container file per chunk, inserting audible silence at every seam (all container formats)

POST /v1/audio/transcriptions: ndjson streaming reports inference-time failures as 200 with an empty body

how to use sesame labs?

Support Stable Audio Open v1.0