Model reference · open weights
xls-r-et is an open-weight audio or speech model from TalTechNLP. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | TalTechNLP |
|---|---|
| Type | Audio & music |
| Task | Speech→text |
| Runs with | transformers |
| Released | 2022-03-02 |
| Popularity | 1k downloads / month |
| Licence | Open weights |
About
This is a XLS-R-300M model facebook/wav2vec2-xls-r-300m finetuned on around 800 hours of diverse Estonian data.
This is a general-purpose Estonian ASR model trained in the Lab of Language Technology at TalTech. It consists of only the CTC-based end-to-end model, no language model is currently provided.
This model is intended for general-purpose speech recognition, such as broadcast conversations, interviews, talks, etc.
TODO
Since this model was trained on mostly broadcast speech and texts from the web, it might have problems correctly decoding the following:
Acoustic training data:
| Type | Amount (h) |
|---|---|
| Broadcast speech | 591 |
| Spontaneous speech | 53 |
| Elderly speech corpus | 53 |
| Talks, lectures | 49 |
| Parliament speeches | 31 |
| Total | 761 |
Finetuned using Fairseq.
| Dataset | WER |
|---|---|
| jutusaated.devset | 7.9 |
| jutusaated.testset | 6.1 |
| Common Voice 6.1 | 12.5 |
| Common Voice 8.0 | 13.4 |
From the published model card. Full card on the HuggingFace links in the sidebar.
Benchmarks
As published on the model card — the maker's own numbers, not measured by AxForge.
| Task | Dataset | Metric | Score |
|---|---|---|---|
| Automatic Speech Recognition | Common Voice | Test WER | 12.520 |
| Automatic Speech Recognition | Common Voice | Test CER | 2.709 |
| Automatic Speech Recognition | Common Voice 8 | Test WER | 13.384 |
| Automatic Speech Recognition | Common Voice 8 | Test CER | 2.982 |
Using it via the API
Once AxForge deploys xls-r-et for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (xls-r-et below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -F model="xls-r-et" -F file=@audio.mp3
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.