Upload folder using huggingface_hub
Browse files
README.md
CHANGED
|
@@ -39,7 +39,12 @@ model-index:
|
|
| 39 |
|
| 40 |
# parakeet-tdt-0.6b-kalenjin
|
| 41 |
|
| 42 |
-
**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
language of Kenya's Rift Valley (~2M speakers). This is an **honest fine-tune** of NVIDIA's
|
| 44 |
[`parakeet-tdt-0.6b-v3`](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) (a 0.6B-parameter
|
| 45 |
FastConformer encoder with a Token-and-Duration Transducer decoder). **It is not trained from
|
|
|
|
| 39 |
|
| 40 |
# parakeet-tdt-0.6b-kalenjin
|
| 41 |
|
| 42 |
+
> ⚠️ **Experimental — not recommended for use.** This Parakeet fine-tune was a research experiment.
|
| 43 |
+
> On real-world Kalenjin audio it is less reliable than the recommended model,
|
| 44 |
+
> [`Tonykip/whisper-kalenjin-v3-turbo`](https://huggingface.co/Tonykip/whisper-kalenjin-v3-turbo)
|
| 45 |
+
> — **use that instead.** This repo is kept for reproducibility and research transparency.
|
| 46 |
+
|
| 47 |
+
**An open automatic speech recognition (ASR) model for Kalenjin** — a Highland Nilotic
|
| 48 |
language of Kenya's Rift Valley (~2M speakers). This is an **honest fine-tune** of NVIDIA's
|
| 49 |
[`parakeet-tdt-0.6b-v3`](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) (a 0.6B-parameter
|
| 50 |
FastConformer encoder with a Token-and-Duration Transducer decoder). **It is not trained from
|