# Temporal Grounding (Human Vocalization) on Audioset Source: [Temporal Grounding (Human Vocalization) on Audioset]() Collection index: [llms.txt]() Data updated: 2026-02-27T11:56:17.237Z Evaluation of temporal grounding performance specifically for human vocalizations on the AudioSet dataset, measured using accuracy tolerances (40ms, 100ms) and Mean Absolute Deviation (MAD). 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 | Temporal Grounding (Human Vocalization) | | Dataset (including any stated split) | Audioset | | Metric | Acc (40ms) | | 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 - [Acc (40ms)](): Higher is better; unit: PERCENT. - [Acc (100ms)](): Higher is better; unit: PERCENT. - [MAD](): Lower is better; unit: SCALAR. ## Recorded results | Method | Acc (40ms) | Result date | Report paper | | --- | --- | --- | --- | | Gemini 2.5 Flash | 6.6 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | Audio Flamingo 3 | 5.8 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | Qwen2.5 7B | 3.9 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | Qwen2.5 3B | 3.6 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | Voxtral 24B | 2.3 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | GPT-4o Audio | 1.9 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | | Voxtral 3B | 1.2 | Not recorded | 698d5e6cc97a0b3e3b5e3cba | ### Gemini 2.5 Flash Acc (40ms): 6.6. Result ID: 69a1865035dfcabb00e79d6c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### Audio Flamingo 3 Acc (40ms): 5.8. Result ID: 69a1865035dfcabb00e79d67. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### Qwen2.5 7B Acc (40ms): 3.9. Result ID: 69a1865035dfcabb00e79d6b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### Qwen2.5 3B Acc (40ms): 3.6. Result ID: 69a1865035dfcabb00e79d6a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### Voxtral 24B Acc (40ms): 2.3. Result ID: 69a1865035dfcabb00e79d69. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### GPT-4o Audio Acc (40ms): 1.9. Result ID: 69a1865035dfcabb00e79d6d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. ### Voxtral 3B Acc (40ms): 1.2. Result ID: 69a1865035dfcabb00e79d68. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Inference Mode":"Zero-shot"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.