patrickvonplaten
commited on
Commit
β’
bdbf2fe
1
Parent(s):
9965f67
Update README.md
Browse files
README.md
CHANGED
@@ -34,13 +34,13 @@ The model can be used directly (without a language model) as follows:
|
|
34 |
import torch
|
35 |
import torchaudio
|
36 |
import librosa
|
37 |
-
from datasets import load_dataset
|
38 |
import MeCab
|
39 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
40 |
|
41 |
# config
|
42 |
wakati = MeCab.Tagger("-Owakati")
|
43 |
-
chars_to_ignore_regex = '[
|
44 |
|
45 |
# load data, processor and model
|
46 |
test_dataset = load_dataset("common_voice", "ja", split="test[:2%]")
|
@@ -73,13 +73,13 @@ The model can be evaluated as follows on the Japanese test data of Common Voice.
|
|
73 |
import torch
|
74 |
import librosa
|
75 |
import torchaudio
|
76 |
-
from datasets import load_dataset
|
77 |
import MeCab
|
78 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
79 |
|
80 |
#config
|
81 |
wakati = MeCab.Tagger("-Owakati")
|
82 |
-
chars_to_ignore_regex = '[
|
83 |
|
84 |
# load data, processor and model
|
85 |
test_dataset = load_dataset("common_voice", "ja", split="test")
|
|
|
34 |
import torch
|
35 |
import torchaudio
|
36 |
import librosa
|
37 |
+
from datasets import load_dataset
|
38 |
import MeCab
|
39 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
40 |
|
41 |
# config
|
42 |
wakati = MeCab.Tagger("-Owakati")
|
43 |
+
chars_to_ignore_regex = '[\\\\\\\\,\\\\\\\\γ\\\\\\\\γ\\\\\\\\οΌ\\\\\\\\γ\\\\\\\\γ\\\\\\\\β¦\\\\\\\\οΌ\\\\\\\\γ»]'
|
44 |
|
45 |
# load data, processor and model
|
46 |
test_dataset = load_dataset("common_voice", "ja", split="test[:2%]")
|
|
|
73 |
import torch
|
74 |
import librosa
|
75 |
import torchaudio
|
76 |
+
from datasets import load_dataset, load_metric
|
77 |
import MeCab
|
78 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
79 |
|
80 |
#config
|
81 |
wakati = MeCab.Tagger("-Owakati")
|
82 |
+
chars_to_ignore_regex = '[\\\\\\\\,\\\\\\\\γ\\\\\\\\γ\\\\\\\\οΌ\\\\\\\\γ\\\\\\\\γ\\\\\\\\β¦\\\\\\\\οΌ\\\\\\\\γ»]'
|
83 |
|
84 |
# load data, processor and model
|
85 |
test_dataset = load_dataset("common_voice", "ja", split="test")
|