Feature: FunASR/SenseVoice for training data annotation

[!NOTE] License and capability clarification (2026-07-14): FunASR is a toolkit, not a single checkpoint. The FunASR and SenseVoice repository source code is MIT; model weights follow each model card. SenseVoiceSmall supports Chinese, Cantonese, English, Japanese, and Korean, and its weights use the linked FunASR Model Open Source License Agreement. Fun-ASR-Nano-2512 is Apache-2.0. Language coverage, punctuation, and performance depend on the selected model and runtime configuration.

Hi! KittenTTS's compact yet high-quality approach is impressive.

For TTS training data preparation (audio transcription/annotation), SenseVoice offers:

• 5x faster than Whisper — process datasets quickly (234M params, non-autoregressive)
• Emotion detection — annotate emotional tone per utterance
• Audio event detection — auto-filter noisy samples
• 50+ languages support

from funasr import AutoModel
model = AutoModel(model="iic/SenseVoiceSmall")
result = model.generate(input="audio.wav")

GPT-SoVITS (58K stars) already uses FunASR for training data annotation.

• FunASR: https://github.com/modelscope/FunASR (16K+ stars)
• SenseVoice: https://github.com/FunAudioLLM/SenseVoice (8K+ stars)