---
language:
  - fon
thumbnail: null
tags:
- automatic-speech-recognition
- CTC
- Attention
- Transformer
- Conformer
- pytorch
- speechbrain
license: apache-2.0
datasets:
- beethogedeon/fongbe-speech
metrics:
- wer
- cer
---

# Fongbe ASR model w/out diacritics

### How to use for inference

```python
from speechbrain.inference.ASR import EncoderASR

asr_model = EncoderASR.from_hparams(
    source="whettenr/asr-fon-without-diacritics",
    savedir="pretrained_models/asr-fongbe-without-diacritics"
)

asr_model.transcribe_file("whettenr/asr-fon-without-diacritics/example.wav")

# expected output:
# huzuhuzu gɔngɔn ɖe ɖo dandan
```

### Details of model
~100M parameters, 12 layer conformer encoder, FFNN decoder 

### Details of training
- pretrained using BEST-RQ on 140 hours
    - FFSTC 2 + beethogedeon/fongbe-speech (~40 hours)
    - cappfm (~100 hours)
- finetuned with CTC loss on training sets of
    - FFSTC 2
    - beethogedeon/fongbe-speech


```
@inproceedings{kponou25_interspeech,
  title     = {{Extending the Fongbe to French Speech Translation Corpus:  resources, models and benchmark}},
  author    = {D. Fortuné Kponou and Salima Mdhaffar and Fréjus A. A. Laleye and Eugène C. Ezin and Yannick Estève},
  year      = {2025},
  booktitle = {{Interspeech 2025}},
  pages     = {4533--4537},
  doi       = {10.21437/Interspeech.2025-1801},
  issn      = {2958-1796},
}
```