arpelarpe commited on
Commit
5e84820
·
1 Parent(s): 9eca7e3

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -10
README.md CHANGED
@@ -3,18 +3,18 @@ license: apache-2.0
3
  ---
4
 
5
  # wav2vec2-base-da-ft-nst
6
- This is a wav2vec2 model for Danish ASR finetuned by Alvenir on the public NST dataset. The model is trained on 16kHz, so make sure your data is the same sample rate.
7
 
8
  The model was trained using fairseq and then converted to huggingface/transformers format.
9
 
10
- Alvenir is always happy to help with your own open-source ASR projects or with customized domain specializations and high performance premium models. ;-)
11
 
12
  ## Usage
13
  ```Python
14
  import soundfile as sf
15
  import torch
16
 
17
- from transformers import Wav2Vec2CTCTokenizer, Wav2Vec2Tokenizer, Wav2Vec2FeatureExtractor, Wav2Vec2Processor, \
18
  Wav2Vec2ForCTC
19
 
20
 
@@ -22,10 +22,6 @@ def get_tokenizer(model_path: str) -> Wav2Vec2CTCTokenizer:
22
  return Wav2Vec2Tokenizer.from_pretrained(model_path)
23
 
24
 
25
- def get_feature_extractor(model_path: str) -> Wav2Vec2FeatureExtractor:
26
- return Wav2Vec2FeatureExtractor.from_pretrained(model_path)
27
-
28
-
29
  def get_processor(model_path: str) -> Wav2Vec2Processor:
30
  return Wav2Vec2Processor.from_pretrained(model_path)
31
 
@@ -55,8 +51,10 @@ print(transcription)
55
 
56
  ```
57
  ## Benchmark results
58
- | Dataset | WER Greddy | WER with Language Model |
 
 
59
  |---------------------|------------|--------------------|
60
  | NST test | 15,8% | 11.9% |
61
- | alvenir-asr-da-eval | 18.2% | 12.1% |
62
- | Common-voice-da | ?? | ?? |
 
3
  ---
4
 
5
  # wav2vec2-base-da-ft-nst
6
+ This the [alvenir wav2vec2 model](https://huggingface.co/Alvenir/wav2vec2-base-da) for Danish ASR finetuned by Alvenir on the public NST dataset. The model is trained on 16kHz, so make sure your data is the same sample rate.
7
 
8
  The model was trained using fairseq and then converted to huggingface/transformers format.
9
 
10
+ Alvenir is always happy to help with your own open-source ASR projects, customized domain specializations or premium models. ;-)
11
 
12
  ## Usage
13
  ```Python
14
  import soundfile as sf
15
  import torch
16
 
17
+ from transformers import Wav2Vec2CTCTokenizer, Wav2Vec2Tokenizer, Wav2Vec2Processor, \
18
  Wav2Vec2ForCTC
19
 
20
 
 
22
  return Wav2Vec2Tokenizer.from_pretrained(model_path)
23
 
24
 
 
 
 
 
25
  def get_processor(model_path: str) -> Wav2Vec2Processor:
26
  return Wav2Vec2Processor.from_pretrained(model_path)
27
 
 
51
 
52
  ```
53
  ## Benchmark results
54
+ This is some benchmark results on the public available datasets in Danish.
55
+
56
+ | Dataset | WER Greedy | WER with Language Model |
57
  |---------------------|------------|--------------------|
58
  | NST test | 15,8% | 11.9% |
59
+ | alvenir-asr-da-eval | 19.0% | 12.1% |
60
+ | common_voice_80 da test | 26,3% | ?? |