patrickvonplaten
commited on
Commit
•
fe26a2e
1
Parent(s):
97cd4e4
Update README.md
Browse files
README.md
CHANGED
@@ -3,11 +3,25 @@ language: zh
|
|
3 |
datasets:
|
4 |
- common_voice
|
5 |
tags:
|
6 |
-
- speech
|
7 |
- audio
|
8 |
- automatic-speech-recognition
|
|
|
9 |
- xlsr-fine-tuning-week
|
10 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
---
|
12 |
|
13 |
## Colab trial with recording or voice file
|
@@ -80,7 +94,7 @@ chars_to_ignore_regex = r"[¥•"#$%&'()*+,-/:;<
|
|
80 |
model = Wav2Vec2ForCTC.from_pretrained(model_name).to(device)
|
81 |
processor = Wav2Vec2Processor.from_pretrained(processor_name)
|
82 |
|
83 |
-
ds = load_dataset("common_voice", 'zh-HK',
|
84 |
|
85 |
resampler = torchaudio.transforms.Resample(orig_freq=48_000, new_freq=16_000)
|
86 |
|
|
|
3 |
datasets:
|
4 |
- common_voice
|
5 |
tags:
|
|
|
6 |
- audio
|
7 |
- automatic-speech-recognition
|
8 |
+
- speech
|
9 |
- xlsr-fine-tuning-week
|
10 |
license: apache-2.0
|
11 |
+
model-index:
|
12 |
+
- name: XLSR Wav2Vec2 Chinese (Hong Kong) by Voidful
|
13 |
+
results:
|
14 |
+
- task:
|
15 |
+
name: Speech Recognition
|
16 |
+
type: automatic-speech-recognition
|
17 |
+
dataset:
|
18 |
+
name: Common Voice zh-HK
|
19 |
+
type: common_voice
|
20 |
+
args: zh-HK
|
21 |
+
metrics:
|
22 |
+
- name: Test CER
|
23 |
+
type: cer
|
24 |
+
value: 76.57
|
25 |
---
|
26 |
|
27 |
## Colab trial with recording or voice file
|
|
|
94 |
model = Wav2Vec2ForCTC.from_pretrained(model_name).to(device)
|
95 |
processor = Wav2Vec2Processor.from_pretrained(processor_name)
|
96 |
|
97 |
+
ds = load_dataset("common_voice", 'zh-HK', split="test")
|
98 |
|
99 |
resampler = torchaudio.transforms.Resample(orig_freq=48_000, new_freq=16_000)
|
100 |
|