Update README.md
Browse files
README.md
CHANGED
@@ -85,34 +85,13 @@ Please notice that we encourage you to read our tutorials and learn more about
|
|
85 |
from speechbrain.pretrained import WhisperASR
|
86 |
|
87 |
asr_model = WhisperASR.from_hparams(source="chaanks/asr-whisper-tiny-sb", savedir="pretrained_models/asr-whisper-tiny-sb")
|
88 |
-
asr_model.transcribe_file("
|
89 |
|
90 |
|
91 |
```
|
92 |
### Inference on GPU
|
93 |
To perform inference on the GPU, add `run_opts={"device":"cuda"}` when calling the `from_hparams` method.
|
94 |
|
95 |
-
### Training
|
96 |
-
The model was trained with SpeechBrain.
|
97 |
-
To train it from scratch follow these steps:
|
98 |
-
1. Clone SpeechBrain:
|
99 |
-
```bash
|
100 |
-
git clone https://github.com/speechbrain/speechbrain/
|
101 |
-
```
|
102 |
-
2. Install it:
|
103 |
-
```bash
|
104 |
-
cd speechbrain
|
105 |
-
pip install -r requirements.txt
|
106 |
-
pip install -e .
|
107 |
-
```
|
108 |
-
|
109 |
-
3. Run Training:
|
110 |
-
```bash
|
111 |
-
cd recipes/CommonVoice/ASR/transformer/
|
112 |
-
python train_with_whisper.py hparams/train_ar_hf_whisper.yaml --data_folder=your_data_folder
|
113 |
-
```
|
114 |
-
|
115 |
-
You can find our training results (models, logs, etc) [here](https://drive.google.com/drive/folders/10mYPYfj9NpDNAa0nO16Zd_K1bIEUOIpx?usp=share_link).
|
116 |
|
117 |
### Limitations
|
118 |
The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
|
|
|
85 |
from speechbrain.pretrained import WhisperASR
|
86 |
|
87 |
asr_model = WhisperASR.from_hparams(source="chaanks/asr-whisper-tiny-sb", savedir="pretrained_models/asr-whisper-tiny-sb")
|
88 |
+
asr_model.transcribe_file("chaanks/asr-whisper-tiny-sb/example.wav")
|
89 |
|
90 |
|
91 |
```
|
92 |
### Inference on GPU
|
93 |
To perform inference on the GPU, add `run_opts={"device":"cuda"}` when calling the `from_hparams` method.
|
94 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
|
96 |
### Limitations
|
97 |
The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.
|