Multi-speaker Speech Transcription and Diarization on AISHELL-4 5-minute long-form (test)
7.73DERSoulX-Transcriber
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SoulX-Transcriberclosed-source=false2026.06 | 7.73 | 14.49 | 17.82 | 3.33 | |
| ViebVoice-ASRclosed-source=false2026.06 | 9.17 | 19.54 | 22.95 | 3.41 | |
| Gemini-3.1-pro-previewclosed-source=true2026.06 | 22.03 | 22.75 | 27.43 | 4.68 | |
| Gemini2.5-Proclosed-source=true2026.06 | 40.87 | 20.26 | 26.31 | 6.05 |