OpenCodePapers

speech-recognition-on-spgispeech

Speech Recognition
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
Click a sortable column header to sort.
PaperCodeWord Error Rate (WER)ModelNameReleaseDate
Introducing Zoom AI Services1.43zoom/scribe_v12026-03-10
Introducing Universal-3 Pro: A new class of speech language model optimized for Voice AI1.52assemblyai/universal-3-pro2026-02-03
1.61OpenMOSS-Team/MOSS-Transcribe-preview-2B2026-06-26
✓ Link1.7nvidia/canary-qwen-2.5b2025-07-17
Training and Inference Efficiency of Encoder-Decoder Speech Models✓ Link1.75nvidia/canary-1b-flash2025-03-07
Voxtral✓ Link1.84mistralai/Voxtral-Small-24B-25072025-07-15
Less is More: Accurate Speech Recognition & Translation without Web-Scale Data✓ Link1.86nvidia/canary-1b2024-02-07
GLM-ASR-Nano: A robust, open-source speech recognition model✓ Link1.86zai-org/GLM-ASR-Nano-25122025-12-09
Streaming Sequence-to-Sequence Learning with Delayed Streams Modeling✓ Link1.87kyutai/stt-2.6b-en2025-06-06
Avalon: ASR for Human–AI Interaction1.88aquavoice/avalon-v1-en2025-08-22
Pulse STT: Fast, Accurate Speech-to-Text1.91smallestai/pulse2026-01-28
Zipformer: A faster and better encoder for automatic speech recognition✓ Link1.91soundsgoodai/Zipformer-transducer-XL-290M2026-05-12
CrisperWhisper: Accurate Timestamps on Verbatim Speech Transcriptions✓ Link1.94nyrahealth/CrisperWhisper2024-08-29
Efficient Sequence Transduction by Jointly Predicting Tokens and Durations✓ Link1.94nvidia/parakeet-tdt-0.6b-v22025-05-01
Canary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST✓ Link2.03nvidia/canary-1b-v22025-08-14
Training and Inference Efficiency of Encoder-Decoder Speech Models✓ Link2.04nvidia/canary-180m-flash2025-03-11
Boson AI Launches Higgs STT 3 Speech-to-Text Model✓ Link2.07bosonai/higgs-audio-v3-stt2026-03-18
Voxtral✓ Link2.15mistralai/Voxtral-Mini-3B-25072025-07-01
Voxtral Realtime✓ Link2.23mistralai/Voxtral-Mini-4B-Realtime-26022026-02-11
Efficient Sequence Transduction by Jointly Predicting Tokens and Durations✓ Link2.26nvidia/parakeet-tdt_ctc-110m2024-09-17
Moonshine v2: Ergodic Streaming Encoder ASR for Latency-Critical Speech Applications✓ Link2.26usefulsensors/moonshine-streaming-medium2026-02-12
OWSM v4: Improving Open Whisper-Style Speech Models via Data Scaling and Cleaning✓ Link2.31espnet/owsm_ctc_v4_1B2025-01-16
Azure Speech at Build 2026: Powering Voice Agents with Real-Time and Life-like Experiences2.31microsoft/azure-speech-06-20262026-06-04
2.35Icefall - zipformer transducer
Data-Efficient On-Policy Distillation for Automatic Speech Recognition✓ Link2.46AutoArk-AI/ARK-ASR-3B2026-06-22
On the Effects of Heterogeneous Data Sources on Speech-to-Text Foundation Models✓ Link2.47espnet/owsm_ctc_v3.2_ft_1B2024-09-24
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link2.56nvidia/parakeet-rnnt-1.1b2023-12-27
LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation✓ Link2.56efficient-speech/lite-whisper-large-v32025-02-26
Qwen3-ASR Technical Report✓ Link2.58Qwen/Qwen3-ASR-1.7B2026-01-28
Robust Knowledge Distillation via Large-Scale Pseudo Labelling✓ Link2.62distil-whisper/distil-large-v3.52024-12-05
LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation✓ Link2.62efficient-speech/lite-whisper-large-v3-acc2025-02-26
Data-Efficient On-Policy Distillation for Automatic Speech Recognition✓ Link2.63AutoArk-AI/ARK-ASR-0.6B2026-05-25
OWSM-CTC: An Open Encoder-Only Speech Foundation Model for Speech Recognition, Translation, and Language Identification✓ Link2.64espnet/owsm_ctc_v3.1_1B2024-02-23
LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation✓ Link2.66efficient-speech/lite-whisper-large-v3-turbo-acc2025-02-26
Stateful Conformer with Cache-based Inference for Streaming Automatic Speech Recognition✓ Link2.69nvidia/nemotron-speech-streaming-en-0.6b2026-03-13
Robust Speech Recognition via Large-Scale Weak Supervision✓ Link2.71openai/whisper-large-v32023-11-06
Efficient Sequence Transduction by Jointly Predicting Tokens and Durations✓ Link2.72nvidia/parakeet-tdt-1.1b2024-01-25
Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs2.73microsoft/Phi-4-multimodal-instruct2025-02-24
Qwen3-ASR Technical Report✓ Link2.74Qwen/Qwen3-ASR-0.6B2026-01-28
Introducing Cohere-transcribe: state-of-the-art speech recognition2.74CohereLabs/cohere-transcribe-03-20262026-03-26
Introducing Resonant-1 and Resonant-1-flash2.77reson8/resonant-12026-04-08
Robust Speech Recognition via Large-Scale Weak Supervision✓ Link2.79openai/whisper-large-v3-turbo2024-10-01
Introducing Resonant-1 and Resonant-1-flash2.8reson8/resonant-1-flash2026-04-08
Ursa 2: Elevating speech recognition across 50+ languages2.81speechmatics/enhanced2024-10-11
Moonshine v2: Ergodic Streaming Encoder ASR for Latency-Critical Speech Applications✓ Link2.85usefulsensors/moonshine-streaming-small2026-02-12
LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation✓ Link2.87efficient-speech/lite-whisper-large-v3-fast2025-02-26
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link2.88nvidia/parakeet-rnnt-0.6b2023-12-28
Introducing Scribe v23.01elevenlabs/scribe_v22026-01-09
Niagara-38m Sets a New Benchmark for Edge Speech Recognition3.08abr-ai/niagara-38m-batch.en2026-04-15
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition3.11parakeet-rnnt-1.1b2023-05-08
Omnilingual ASR: Open-Source Multilingual Speech Recognition for 1600+ Languages✓ Link3.22facebook/omniASR-LLM-7B-v22025-12-12
NLE: Non-autoregressive LLM-based ASR by Transcript Editing✓ Link3.23ibm-granite/granite-speech-4.1-2b-nar2026-04-01
Boson AI Launches Higgs STT 3 Speech-to-Text Model✓ Link3.23bosonai/higgs-audio-v3-8b-stt-v22026-04-27
Stateful Conformer with Cache-based Inference for Streaming Automatic Speech Recognition✓ Link3.27nvidia/nemotron-3.5-asr-streaming-0.6b2026-06-04
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link3.47nvidia/parakeet-ctc-0.6b2023-12-28
VibeVoice-ASR Technical Report✓ Link3.47microsoft/VibeVoice-ASR-HF2026-03-02
✓ Link3.47ibm-granite/granite-speech-4.1-2b2026-04-29
Granite-speech: open-source speech-aware LLMs with strong English ASR capabilities✓ Link3.5ibm-granite/granite-speech-3.3-2b2025-06-19
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link3.53nvidia/parakeet-ctc-1.1b2023-12-28
Granite-speech: open-source speech-aware LLMs with strong English ASR capabilities✓ Link3.54ibm-granite/granite-speech-3.3-8b2025-06-19
✓ Link3.55ibm-granite/granite-4.0-1b-speech2026-03-06
Canary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST✓ Link3.63nvidia/parakeet-tdt-0.6b-v32025-08-14
Loquacious Set: 25,000 Hours of Transcribed and Diverse English Speech Recognition Data for Research and Commercial Use✓ Link3.69speechbrain/asr-conformer-largescaleasr2025-02-06
Niagara-38m Sets a New Benchmark for Edge Speech Recognition3.86abr-ai/niagara-19m-batch.en2026-04-15
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link4.03nvidia/stt_en_fastconformer_transducer_large2023-06-08
Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition✓ Link4.25nvidia/stt_en_fastconformer_ctc_large2023-06-08
Conformer: Convolution-augmented Transformer for Speech Recognition✓ Link4.79nvidia/stt_en_conformer_ctc_large2022-04-09
Moonshine: Speech Recognition for Live Transcription and Voice Commands✓ Link4.96usefulsensors/moonshine-base2024-10-21
Omnilingual ASR: Open-Source Multilingual Speech Recognition for 1600+ Languages✓ Link5.16facebook/omniASR-CTC-7B-v22025-12-12
SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition✓ Link5.7Conformer2021-04-05
Moonshine v2: Ergodic Streaming Encoder ASR for Latency-Critical Speech Applications✓ Link5.71usefulsensors/moonshine-streaming-tiny2026-02-12
Conformer: Convolution-augmented Transformer for Speech Recognition✓ Link5.96nvidia/stt_en_conformer_transducer_small2022-06-01
Moonshine: Speech Recognition for Live Transcription and Voice Commands✓ Link6.88usefulsensors/moonshine-tiny2024-10-21
Conformer: Convolution-augmented Transformer for Speech Recognition✓ Link7.01nvidia/stt_en_conformer_ctc_small2023-06-12
Scaling Speech Technology to 1,000+ Languages✓ Link7.49facebook/mms-1b-all2023-05-27
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations✓ Link7.81facebook/wav2vec2-large-960h-lv60-self2020-06-20
data2vec: A General Framework for Self-supervised Learning in Speech, Vision and Language✓ Link8.35facebook/data2vec-audio-large-960h2022-04-02
fairseq S2T: Fast Speech-to-Text Modeling with fairseq✓ Link8.37facebook/wav2vec2-conformer-rel-pos-large-960h-ft2022-04-18
fairseq S2T: Fast Speech-to-Text Modeling with fairseq✓ Link8.61facebook/wav2vec2-conformer-rope-large-960h-ft2022-04-18
SpeechBrain: A General-Purpose Speech Toolkit✓ Link8.62speechbrain/asr-wav2vec2-librispeech2022-06-05
HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units✓ Link8.66facebook/hubert-xlarge-ls960-ft2022-03-02
HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units✓ Link8.95facebook/hubert-large-ls960-ft2022-03-02
Robust wav2vec 2.0: Analyzing Domain Shift in Self-Supervised Pre-Training✓ Link9.27facebook/wav2vec2-large-robust-ft-libri-960h2021-04-02
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations✓ Link13.31facebook/wav2vec2-large-960h2020-06-20
data2vec: A General Framework for Self-supervised Learning in Speech, Vision and Language✓ Link16.28facebook/data2vec-audio-base-960h2022-03-02
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations✓ Link18.44facebook/wav2vec2-base-960h2022-03-02
Check out how to contribute new results. Then edit this file.