# Lipreading on GRID (unseen speakers)

Source: [Lipreading on GRID (unseen speakers)](<https://www.sota2.com/research/sota/lipreading-on-grid-unseen-speakers>)

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

Data updated: 2026-02-26T06:08:23.793Z

Evaluation of Lipreading performance on the GRID dataset, specifically focusing on unseen speakers, typically measured using Character Error Rate (CER) or Word Error Rate (WER).

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 | Lipreading |
| Dataset (including any stated split) | GRID (unseen speakers) |
| Metric | WER |
| 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

- [WER](<https://www.sota2.com/research/sota/lipreading-on-grid-unseen-speakers.md?metric=WER>): Lower is better; unit: PERCENT.

- [CER](<https://www.sota2.com/research/sota/lipreading-on-grid-unseen-speakers.md?metric=CER>): Lower is better; unit: SCALAR.

## Recorded results

| Method | WER | Result date | Report paper |
| --- | --- | --- | --- |
| LipNet | 11.4 | Not recorded | 68fa63091e209131ce9fc909 |
| Baseline-NoLM | 13.6 | Not recorded | 68fa63091e209131ce9fc909 |
| Baseline-2D | 26.7 | Not recorded | 68fa63091e209131ce9fc909 |
| Hearing-Impaired Person (avg) | 47.7 | Not recorded | 68fa63091e209131ce9fc909 |
| Baseline-LSTM | 52.8 | Not recorded | 68fa63091e209131ce9fc909 |

### LipNet

WER: 11.4. Result ID: 698be7c7ecace47ef559f36e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

[Implementation](<https://youtube.com/playlist?list=PLXkuFIFnXUAPIrXKgtIpctv2NuSo7xw3k>)

Recorded evaluation configuration: {"Search strategy":"CTC beam search"}. This record does not assert independent reproduction.

### Baseline-NoLM

WER: 13.6. Result ID: 698be7c7ecace47ef559f36d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

Recorded evaluation configuration: {"Language model":"Disabled"}. This record does not assert independent reproduction.

### Baseline-2D

WER: 26.7. Result ID: 698be7c7ecace47ef559f36c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

Recorded evaluation configuration: {"Architecture":"Spatial-only convolutions"}. This record does not assert independent reproduction.

### Hearing-Impaired Person (avg)

WER: 47.7. Result ID: 698be7c7ecace47ef559f36a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

Recorded evaluation configuration: {"Participant group":"Three members of the Oxford Students' Disability Community"}. This record does not assert independent reproduction.

### Baseline-LSTM

WER: 52.8. Result ID: 698be7c7ecace47ef559f36b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa63091e209131ce9fc909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62441e209131ce9f8527.md>)

Recorded evaluation configuration: {"Architecture":"Replication of Wand et al., 2016"}. This record does not assert independent reproduction.

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