# Speaker Diarization on NIST RT-03 (LDC2007S10) (eval)

Source: [Speaker Diarization on NIST RT-03 (LDC2007S10) (eval)](<https://www.sota2.com/research/sota/speaker-diarization-on-nist-rt-03-ldc2007s10-eval>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T06:04:32.709Z

Evaluation of Speaker Diarization performance on the evaluation split of the NIST RT-03 (LDC2007S10) dataset, measured using DER and Speaker Error Rate.

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 | Speaker Diarization |
| Dataset (including any stated split) | NIST RT-03 (LDC2007S10) (eval) |
| Metric | Speaker Error Rate |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 5 of 5 |
| 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

- [Speaker Error Rate](<https://www.sota2.com/research/sota/speaker-diarization-on-nist-rt-03-ldc2007s10-eval.md?metric=Speaker%20Error%20Rate>): Lower is better; unit: PERCENT.

- [DER](<https://www.sota2.com/research/sota/speaker-diarization-on-nist-rt-03-ldc2007s10-eval.md?metric=DER>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Speaker Error Rate | Result date | Report paper |
| --- | --- | --- | --- |
| COS+B-SC | 0.88 | Not recorded | 68fa67d01e209131cea189a2 |
| PLDA+AHC | 1.73 | Not recorded | 68fa67d01e209131cea189a2 |
| COS+NME-SC | 2.21 | Not recorded | 68fa67d01e209131cea189a2 |
| COS+AHC | 5.66 | Not recorded | 68fa67d01e209131cea189a2 |
| COS+NJW-SC | 6.56 | Not recorded | 68fa67d01e209131cea189a2 |

### COS+B-SC

Speaker Error Rate: 0.88. Result ID: 698d47448b517eb433204f7f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

[Implementation](<https://www.sota2.com/github.com/tango4j/Auto-Tuning-Spectral-Clustering>)

Recorded evaluation configuration: {"SAD Type":"Oracle","Speaker Count Estimation":"Estimated"}. This record does not assert independent reproduction.

### PLDA+AHC

Speaker Error Rate: 1.73. Result ID: 698d47448b517eb433204f7e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

Recorded evaluation configuration: {"SAD Type":"Oracle","Speaker Count Estimation":"Estimated"}. This record does not assert independent reproduction.

### COS+NME-SC

Speaker Error Rate: 2.21. Result ID: 698d47448b517eb433204f80.

[Reporting paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

[Implementation](<https://www.sota2.com/github.com/tango4j/Auto-Tuning-Spectral-Clustering>)

Recorded evaluation configuration: {"SAD Type":"Oracle","Speaker Count Estimation":"Estimated"}. This record does not assert independent reproduction.

### COS+AHC

Speaker Error Rate: 5.66. Result ID: 698d47448b517eb433204f7d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

Recorded evaluation configuration: {"SAD Type":"Oracle","Speaker Count Estimation":"Estimated"}. This record does not assert independent reproduction.

### COS+NJW-SC

Speaker Error Rate: 6.56. Result ID: 698d47448b517eb433204f7c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa67d01e209131cea189a2.md>)

Recorded evaluation configuration: {"SAD Type":"Oracle","Speaker Count Estimation":"Estimated"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.