---
language:
- tw
- ak
license: apache-2.0
library_name: transformers
pipeline_tag: automatic-speech-recognition
base_model: KhayaAI/w2v-bert-ada_ewe_fat_fra_gaa_nzi_twi_en
tags:
- ak
- akan
- asante-twi
- asr
- ctc
- dondo
- ghana
- ghana-health-ai
- serendepify
- tw
- twi
- w2v-bert
datasets:
- google/WaxalNLP
model-index:
- name: gha-dondo-w2v-bert-twi-v1
  results:
  - task:
      type: automatic-speech-recognition
    dataset:
      type: google/WaxalNLP
      name: google/WaxalNLP
    metrics:
    - type: val_wer
      value: 0.357730
      name: VAL_WER
    - type: val_cer
      value: 0.121899
      name: VAL_CER
---

# teckedd/gha-dondo-w2v-bert-twi-v1

DONDO w2v-BERT CTC fine-tune trial for Ghana Health AI (`dondo-waxal-twi-v1`). Promotion candidate: False.

- **Base model:** `KhayaAI/w2v-bert-ada_ewe_fat_fra_gaa_nzi_twi_en`
- **Task:** `automatic-speech-recognition`
- **Languages:** `tw`, `ak`
- **Card generated:** 2026-08-12 (UTC)
- **Product:** [ghanahealth.serendepify.com](https://ghanahealth.serendepify.com)

> Not a medical device. Outputs support community health guidance only.


## Metrics

| Metric | Value |
|--------|-------|
| `val_wer` | 0.3577 |
| `val_cer` | 0.1219 |


## Training data

- `google/WaxalNLP`


## Intended use

- In-product ASR / TTS / chat for Twi (Akan) and English health conversations in Ghana.
- Research on low-resource Ghanaian language speech and health dialogue.

## Out of scope

- Clinical diagnosis or autonomous medical decisions.
- Claiming near-native quality without reporting held-out WER/CER or human A/B scores.

## How to load

```python
from transformers import pipeline  # or AutoModel + processor per task
# repo: teckedd/gha-dondo-w2v-bert-twi-v1
```

## Citation

If you use this checkpoint, please credit **Ghana Health AI / Serendepify** and the upstream base model authors plus any listed datasets.


## DONDO language prefix

This checkpoint follows DONDO's language-conditioned CTC setup. For Asante Twi, prepend
language id `2` to acoustic features before decoding.

## Promotion gate

- Current v6 Whisper beam=5 WER: `0.3044`
- This validation WER: `0.3577304964539007`
- Promote: **False**