Automatic Speech Recognition
Transformers
Safetensors
Welsh
English
whisper
welsh
cymraeg
speech-recognition
translation
Instructions to use techiaith/whisper-large-ft-cy-en with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use techiaith/whisper-large-ft-cy-en with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="techiaith/whisper-large-ft-cy-en")# Load model directly from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq processor = AutoProcessor.from_pretrained("techiaith/whisper-large-ft-cy-en") model = AutoModelForSpeechSeq2Seq.from_pretrained("techiaith/whisper-large-ft-cy-en") - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,17 +1,28 @@
|
|
| 1 |
---
|
| 2 |
language:
|
| 3 |
-
|
| 4 |
-
|
| 5 |
license: apache-2.0
|
| 6 |
library_name: transformers
|
| 7 |
pipeline_tag: automatic-speech-recognition
|
| 8 |
tags:
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
base_model: openai/whisper-large-v2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
---
|
| 16 |
|
| 17 |
# Whisper Large — Welsh & English (techiaith/whisper-large-ft-cy-en)
|
|
@@ -54,14 +65,14 @@ Total training data: **~177 hours** across **153,066 clips**.
|
|
| 54 |
|
| 55 |
| Dataset | Language | Duration | Clips | Description |
|
| 56 |
|---------|----------|----------|-------|-------------|
|
| 57 |
-
| [
|
| 58 |
| [techiaith/corpws-clllc-wlga](https://huggingface.co/datasets/techiaith/corpws-clllc-wlga) | Welsh | 32:59h | 26,216 | Local government meetings |
|
| 59 |
| [cymen-arfor/lleisiau-arfor](https://huggingface.co/datasets/cymen-arfor/lleisiau-arfor) | Welsh | 33:54h | 33,614 | Spontaneous conversational speech |
|
| 60 |
| [techiaith/commonvoice_23_0_cy](https://huggingface.co/datasets/techiaith/commonvoice_23_0_cy) | Welsh | 31:11h | 20,018 | Read speech (CommonVoice 23.0) |
|
| 61 |
| [techiaith/commonvoice_vad_cy](https://huggingface.co/datasets/techiaith/commonvoice_vad_cy) | Welsh | 3:27h | 8,209 | VAD-segmented clips |
|
| 62 |
| [techiaith/commonvoice_23_0_en__GB_IE](https://huggingface.co/datasets/techiaith/commonvoice_23_0_en__GB_IE) | English | 22:26h | 16,440 | Read speech, UK/Irish accents (10% sample) |
|
| 63 |
|
| 64 |
-
Validation: [
|
| 65 |
|
| 66 |
## Training Configuration
|
| 67 |
|
|
@@ -108,4 +119,4 @@ A CTranslate2 (int8 quantised) version is available at [techiaith/whisper-large-
|
|
| 108 |
|
| 109 |
Developed by [Uned Technolegau Iaith, Prifysgol Bangor / Language Technologies Unit, Bangor University](https://techiaith.bangor.ac.uk/).
|
| 110 |
|
| 111 |
-
Funded by the Welsh Government.
|
|
|
|
| 1 |
---
|
| 2 |
language:
|
| 3 |
+
- cy
|
| 4 |
+
- en
|
| 5 |
license: apache-2.0
|
| 6 |
library_name: transformers
|
| 7 |
pipeline_tag: automatic-speech-recognition
|
| 8 |
tags:
|
| 9 |
+
- whisper
|
| 10 |
+
- welsh
|
| 11 |
+
- cymraeg
|
| 12 |
+
- speech-recognition
|
| 13 |
+
- translation
|
| 14 |
base_model: openai/whisper-large-v2
|
| 15 |
+
datasets:
|
| 16 |
+
- techiaith/banc-trawsgrifiadau-bangor
|
| 17 |
+
- techiaith/corpws-clllc-wlga
|
| 18 |
+
- cymen-arfor/lleisiau-arfor
|
| 19 |
+
- techiaith/commonvoice_23_0_cy
|
| 20 |
+
- techiaith/commonvoice_vad_cy
|
| 21 |
+
- techiaith/commonvoice_23_0_en__GB_IE
|
| 22 |
+
- techiaith/commonvoice-23-0-cy-en
|
| 23 |
+
metrics:
|
| 24 |
+
- wer
|
| 25 |
+
- cer
|
| 26 |
---
|
| 27 |
|
| 28 |
# Whisper Large — Welsh & English (techiaith/whisper-large-ft-cy-en)
|
|
|
|
| 65 |
|
| 66 |
| Dataset | Language | Duration | Clips | Description |
|
| 67 |
|---------|----------|----------|-------|-------------|
|
| 68 |
+
| [techiaith/banc-trawsgrifiadau-bangor](https://huggingface.co/datasets/techiaith/banc-trawsgrifiadau-bangor) | Welsh | 52:45h | 48,569 | Mixed spontaneous & read speech |
|
| 69 |
| [techiaith/corpws-clllc-wlga](https://huggingface.co/datasets/techiaith/corpws-clllc-wlga) | Welsh | 32:59h | 26,216 | Local government meetings |
|
| 70 |
| [cymen-arfor/lleisiau-arfor](https://huggingface.co/datasets/cymen-arfor/lleisiau-arfor) | Welsh | 33:54h | 33,614 | Spontaneous conversational speech |
|
| 71 |
| [techiaith/commonvoice_23_0_cy](https://huggingface.co/datasets/techiaith/commonvoice_23_0_cy) | Welsh | 31:11h | 20,018 | Read speech (CommonVoice 23.0) |
|
| 72 |
| [techiaith/commonvoice_vad_cy](https://huggingface.co/datasets/techiaith/commonvoice_vad_cy) | Welsh | 3:27h | 8,209 | VAD-segmented clips |
|
| 73 |
| [techiaith/commonvoice_23_0_en__GB_IE](https://huggingface.co/datasets/techiaith/commonvoice_23_0_en__GB_IE) | English | 22:26h | 16,440 | Read speech, UK/Irish accents (10% sample) |
|
| 74 |
|
| 75 |
+
Validation: [techiaith/banc-trawsgrifiadau-bangor](https://huggingface.co/datasets/DewiBrynJones/banc-trawsgrifiadau-bangor) validation split (4:00h, 3,895 clips).
|
| 76 |
|
| 77 |
## Training Configuration
|
| 78 |
|
|
|
|
| 119 |
|
| 120 |
Developed by [Uned Technolegau Iaith, Prifysgol Bangor / Language Technologies Unit, Bangor University](https://techiaith.bangor.ac.uk/).
|
| 121 |
|
| 122 |
+
Funded by the Welsh Government.
|