# Audio Understanding on Big Bench Audio Source: [Audio Understanding on Big Bench Audio]() Collection index: [llms.txt]() Data updated: 2026-06-09T10:34:17.511Z Evaluation of model performance on the Big Bench Audio dataset focusing on general Audio Understanding capabilities, measured by Audio accuracy percentage. 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 | Audio Understanding | | Dataset (including any stated split) | Big Bench Audio | | Metric | Audio Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 7 of 7 | | 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 - [Audio Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Audio Accuracy | Result date | Report paper | | --- | --- | --- | --- | | DuplexOmni | 77.2 | Not recorded | 6a279e5690966ade919bf9e4 | | Gemini-3.1-Flash-Lite | 58.9 | Not recorded | 6a279e5690966ade919bf9e4 | | Gemini-3.1-Flash-Live | 57.9 | Not recorded | 6a279e5690966ade919bf9e4 | | Doubao | 47.9 | Not recorded | 6a279e5690966ade919bf9e4 | | MiniCPM-o 4.5 | 36 | Not recorded | 6a279e5690966ade919bf9e4 | | Qwen3.5-Omni-Realtime-Flash | 35 | Not recorded | 6a279e5690966ade919bf9e4 | | Qwen3-Omni-Realtime-Flash | 32.1 | Not recorded | 6a279e5690966ade919bf9e4 | ### DuplexOmni Audio Accuracy: 77.2. Result ID: 6a27eb1bfbc373592c256132. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"0.506"}. This record does not assert independent reproduction. ### Gemini-3.1-Flash-Lite Audio Accuracy: 58.9. Result ID: 6a27eb1bfbc373592c256138. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode"}. This record does not assert independent reproduction. ### Gemini-3.1-Flash-Live Audio Accuracy: 57.9. Result ID: 6a27eb1bfbc373592c256137. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"2.57"}. This record does not assert independent reproduction. ### Doubao Audio Accuracy: 47.9. Result ID: 6a27eb1bfbc373592c256134. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"1.82"}. This record does not assert independent reproduction. ### MiniCPM-o 4.5 Audio Accuracy: 36. Result ID: 6a27eb1bfbc373592c256133. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"0.502"}. This record does not assert independent reproduction. ### Qwen3.5-Omni-Realtime-Flash Audio Accuracy: 35. Result ID: 6a27eb1bfbc373592c256136. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"1.25"}. This record does not assert independent reproduction. ### Qwen3-Omni-Realtime-Flash Audio Accuracy: 32.1. Result ID: 6a27eb1bfbc373592c256135. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"real-time mode","Latency (s)":"1.28"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.