Update README.md
Browse files
README.md
CHANGED
@@ -36,13 +36,16 @@ The model is fine-tuned on the [LibriMix dataset](https://github.com/JorisCos/Li
|
|
36 |
from transformers import Wav2Vec2FeatureExtractor, UniSpeechSatForAudioFrameClassification
|
37 |
from datasets import load_dataset
|
38 |
import torch
|
|
|
39 |
dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
|
40 |
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained('microsoft/wavlm-base-plus-sd')
|
41 |
model = UniSpeechSatForAudioFrameClassification.from_pretrained('microsoft/wavlm-base-plus-sd')
|
|
|
42 |
# audio file is decoded on the fly
|
43 |
inputs = feature_extractor(dataset[0]["audio"]["array"], return_tensors="pt")
|
44 |
logits = model(**inputs).logits
|
45 |
probabilities = torch.sigmoid(logits[0])
|
|
|
46 |
# labels is a one-hot array of shape (num_frames, num_speakers)
|
47 |
labels = (probabilities > 0.5).long()
|
48 |
```
|
|
|
36 |
from transformers import Wav2Vec2FeatureExtractor, UniSpeechSatForAudioFrameClassification
|
37 |
from datasets import load_dataset
|
38 |
import torch
|
39 |
+
|
40 |
dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
|
41 |
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained('microsoft/wavlm-base-plus-sd')
|
42 |
model = UniSpeechSatForAudioFrameClassification.from_pretrained('microsoft/wavlm-base-plus-sd')
|
43 |
+
|
44 |
# audio file is decoded on the fly
|
45 |
inputs = feature_extractor(dataset[0]["audio"]["array"], return_tensors="pt")
|
46 |
logits = model(**inputs).logits
|
47 |
probabilities = torch.sigmoid(logits[0])
|
48 |
+
|
49 |
# labels is a one-hot array of shape (num_frames, num_speakers)
|
50 |
labels = (probabilities > 0.5).long()
|
51 |
```
|