# Automatic Speech Recognition on SpeechIO Source: [Automatic Speech Recognition on SpeechIO]() Collection index: [llms.txt]() Data updated: 2026-04-21T12:37:51.925Z Evaluation of Automatic Speech Recognition performance on the SpeechIO dataset, typically measured using Character Error Rate (CER). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Automatic Speech Recognition | | Dataset (including any stated split) | SpeechIO | | Metric | Character Error Rate (CER) | | Unit | PERCENT | | Direction | Lower is better | | Results loaded | 14 of 14 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Character Error Rate (CER)](): Lower is better; unit: PERCENT. ## Recorded results | Method | Character Error Rate (CER) | Result date | Report paper | | --- | --- | --- | --- | | FireRedASR2S LLM | 2.2 | Not recorded | 69e704de7d0172140ec7f08e | | Qwen3-Omni Instruct | 2.33 | Not recorded | 69e704de7d0172140ec7f08e | | Qwen3-ASR 1.7B | 2.55 | Not recorded | 69e704de7d0172140ec7f08e | | NIM4-ASR | 2.61 | Not recorded | 69e704de7d0172140ec7f08e | | Fun-ASR Nano | 2.78 | Not recorded | 69e704de7d0172140ec7f08e | | NIM4-ASR | 2.84 | Not recorded | 69e704de7d0172140ec7f08e | | Qwen3-ASR-1.7B | 2.88 | Not recorded | 697c3afda8b01e76cc8da424 | | Doubao-ASR | 2.93 | Not recorded | 697c3afda8b01e76cc8da424 | | GLM-ASR Nano | 3.17 | Not recorded | 69e704de7d0172140ec7f08e | | Step-Audio2 Mini | 3.41 | Not recorded | 69e704de7d0172140ec7f08e | | Qwen3-ASR-0.6B | 3.44 | Not recorded | 697c3afda8b01e76cc8da424 | | Gemini-2.5-Pro | 5.3 | Not recorded | 697c3afda8b01e76cc8da424 | | Whisper-large-v3 | 7.56 | Not recorded | 697c3afda8b01e76cc8da424 | | GPT-4o-Transcribe | 12.86 | Not recorded | 697c3afda8b01e76cc8da424 | ### FireRedASR2S LLM Character Error Rate (CER): 2.2. Result ID: 69e76f8d47f34ce924cf169f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Size":"8B+","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### Qwen3-Omni Instruct Character Error Rate (CER): 2.33. Result ID: 69e76f8d47f34ce924cf16a1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"30B-A3B","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### Qwen3-ASR 1.7B Character Error Rate (CER): 2.55. Result ID: 69e76f8d47f34ce924cf169e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"2.0B","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### NIM4-ASR Character Error Rate (CER): 2.61. Result ID: 69e76f8d47f34ce924cf16a2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"2.3B","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### Fun-ASR Nano Character Error Rate (CER): 2.78. Result ID: 69e76f8d47f34ce924cf169c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"0.8B","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### NIM4-ASR Character Error Rate (CER): 2.84. Result ID: 69e76f8d47f34ce924cf16a3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"2.3B","Inference Mode":"Stream"}. This record does not assert independent reproduction. ### Qwen3-ASR-1.7B Character Error Rate (CER): 2.88. Result ID: 699664bb5698665ec4aaf188. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Parameters":"1.7B"}. This record does not assert independent reproduction. ### Doubao-ASR Character Error Rate (CER): 2.93. Result ID: 699664bb5698665ec4aaf185. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GLM-ASR Nano Character Error Rate (CER): 3.17. Result ID: 69e76f8d47f34ce924cf169d. [Reporting paper]() Recorded evaluation configuration: {"Model Size":"1.5B","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### Step-Audio2 Mini Character Error Rate (CER): 3.41. Result ID: 69e76f8d47f34ce924cf16a0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size":"8B+","Inference Mode":"Offline"}. This record does not assert independent reproduction. ### Qwen3-ASR-0.6B Character Error Rate (CER): 3.44. Result ID: 699664bb5698665ec4aaf187. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Parameters":"0.6B"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Character Error Rate (CER): 5.3. Result ID: 699664bb5698665ec4aaf184. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Whisper-large-v3 Character Error Rate (CER): 7.56. Result ID: 699664bb5698665ec4aaf186. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-4o-Transcribe Character Error Rate (CER): 12.86. Result ID: 699664bb5698665ec4aaf183. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.