# Zero-shot Text-to-Speech Synthesis on LibriSpeech 3s audio prompt (test-clean)

Source: [Zero-shot Text-to-Speech Synthesis on LibriSpeech 3s audio prompt (test-clean)](<https://www.sota2.com/research/sota/zero-shot-text-to-speech-synthesis-on-librispeech-3s-audio-prompt-test-clean>)

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

Data updated: 2026-02-26T06:00:51.692Z

Evaluation of Zero-shot Text-to-Speech synthesis performance on the LibriSpeech test-clean split, specifically using 3-second audio prompts, measured by accuracy (WER, SIM-O) and efficiency (NFE, RTF).

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 | Zero-shot Text-to-Speech Synthesis |
| Dataset (including any stated split) | LibriSpeech 3s audio prompt (test-clean) |
| Metric | RTF |
| Unit | SCALAR |
| 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

- [RTF](<https://www.sota2.com/research/sota/zero-shot-text-to-speech-synthesis-on-librispeech-3s-audio-prompt-test-clean.md?metric=RTF>): Lower is better; unit: SCALAR.

- [NFE](<https://www.sota2.com/research/sota/zero-shot-text-to-speech-synthesis-on-librispeech-3s-audio-prompt-test-clean.md?metric=NFE>): Lower is better; unit: SCALAR.

- [WER](<https://www.sota2.com/research/sota/zero-shot-text-to-speech-synthesis-on-librispeech-3s-audio-prompt-test-clean.md?metric=WER>): Lower is better; unit: PERCENT.

- [SIM-O](<https://www.sota2.com/research/sota/zero-shot-text-to-speech-synthesis-on-librispeech-3s-audio-prompt-test-clean.md?metric=SIM-O>): Higher is better; unit: PERCENT.

## Recorded results

| Method | RTF | Result date | Report paper |
| --- | --- | --- | --- |
| OZSpeech | 0.26 | Not recorded | 68fa789b1e209131cea794b4 |
| F5-TTS | 0.7 | Not recorded | 68fa789b1e209131cea794b4 |
| VALL-E | 0.86 | Not recorded | 68fa789b1e209131cea794b4 |
| NaturalSpeech 2 | 1.66 | Not recorded | 68fa789b1e209131cea794b4 |
| VoiceCraft | 1.7 | Not recorded | 68fa789b1e209131cea794b4 |

### OZSpeech

RTF: 0.26. Result ID: 699594a73badfaffe68f424e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Implementation](<https://ozspeech.github.io/OZSpeech_Web/>)

Recorded evaluation configuration: {"Trainable Parameters":"145M","Frozen Parameters":"102M","Codec":"FACodec"}. This record does not assert independent reproduction.

### F5-TTS

RTF: 0.7. Result ID: 699594a73badfaffe68f424a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74cc1e209131cea63eaf.md>)

Recorded evaluation configuration: {"Trainable Parameters":"336M","Frozen Parameters":"13.5M","Vocoder":"Vocos"}. This record does not assert independent reproduction.

### VALL-E

RTF: 0.86. Result ID: 699594a73badfaffe68f424d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e5c1e209131cea3f57e.md>)

Recorded evaluation configuration: {"Trainable Parameters":"594M","Frozen Parameters":"104M","Tokenizer":"SpeechTokenizer"}. This record does not assert independent reproduction.

### NaturalSpeech 2

RTF: 1.66. Result ID: 699594a73badfaffe68f424c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f931e209131cea465bb.md>)

Recorded evaluation configuration: {"Trainable Parameters":"378M","Frozen Parameters":"14M","Codec":"EnCodec"}. This record does not assert independent reproduction.

### VoiceCraft

RTF: 1.7. Result ID: 699594a73badfaffe68f424b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789b1e209131cea794b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73a61e209131cea5daf3.md>)

Recorded evaluation configuration: {"Trainable Parameters":"830M","Frozen Parameters":"14M","Codec":"EnCodec"}. This record does not assert independent reproduction.

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