Qwen/Qwen3-ASR-1.7B

Automatic Speech Recognition • 2B • Updated Jan 30 • 1.45M • • 1.21k

Note Multilingual ASR: 30 languages and 22 Chinese dialects. Offline and streaming inference; streaming uses the vLLM backend. Apache-2.0.

nvidia/parakeet-tdt-0.6b-v3

Automatic Speech Recognition • 0.6B • Updated Aug 5 • 583k • 1.24k

Note Compact 0.6B multilingual ASR covering 25 European languages. Useful transcription baseline in the NeMo/Transformers ecosystem. CC-BY-4.0.

openai/whisper-large-v3-turbo

Automatic Speech Recognition • 0.8B • Updated Oct 4, 2024 • 5.78M • • 3.49k

Note Widely supported multilingual transcription baseline. Faster variant of Whisper large-v3. MIT.

kyutai/stt-1b-en_fr

Automatic Speech Recognition • 1.0B • Updated Nov 18, 2025 • 138

Note Native streaming ASR for English and French. Produces punctuated transcripts while audio arrives. CC-BY-4.0.

nvidia/nemotron-3.5-asr-streaming-0.6b

Automatic Speech Recognition • 0.6B • Updated Sep 10 • 838k • • 1.24k

Note Native cache-aware streaming ASR with configurable chunk sizes and punctuation. Multilingual coverage is tiered; some locales require fine-tuning. OpenMDW-1.1.

mistralai/Voxtral-Mini-4B-Realtime-2602

Automatic Speech Recognition • 4B • Updated Mar 11 • 1.94M • 1.05k

Note Native streaming multilingual ASR with a causal audio encoder and configurable transcription delay. Useful for latency/accuracy comparisons. Apache-2.0.

nvidia/canary-qwen-2.5b

Automatic Speech Recognition • 3B • Updated Apr 21 • 28.6k • 474

Note English ASR with punctuation and capitalization. Separate LLM mode can process the transcript; raw-audio understanding is not retained in that mode. CC-BY-4.0.

FunAudioLLM/SenseVoiceSmall

Automatic Speech Recognition • Updated Jun 20 • 23.9k • 503

Note Non-autoregressive ASR with emotion recognition and audio-event tags. Useful for richer transcription than words alone. Custom model license.

FireRedTeam/FireRedASR2-AED

Automatic Speech Recognition • Updated Mar 13 • 1.33k • 37

Note Mandarin, Chinese dialects/accents, English and code-switching; speech and singing transcription. ASR component of FireRedASR2S. Apache-2.0.

distil-whisper/distil-large-v3

Automatic Speech Recognition • 0.8B • Updated Apr 21 • 737k • 384

Note Distilled Whisper large-v3 for English transcription, including long-form audio. Useful speed-oriented Whisper baseline. MIT.