bweng commited on
Commit
0cd1e02
·
verified ·
1 Parent(s): 883ca64

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +151 -0
README.md CHANGED
@@ -1,3 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  # Parakeet-TDT-CTC-110M CoreML
2
 
3
  NVIDIA's Parakeet-TDT-CTC-110M model converted to CoreML format for efficient inference on Apple Silicon.
 
1
+ ---
2
+ language:
3
+ - en
4
+ library_name: nemo
5
+ datasets:
6
+ - librispeech_asr
7
+ - fisher_corpus
8
+ - mozilla-foundation/common_voice_8_0
9
+ - National-Singapore-Corpus-Part-1
10
+ - vctk
11
+ - voxpopuli
12
+ - europarl
13
+ - multilingual_librispeech
14
+
15
+ thumbnail: null
16
+ tags:
17
+ - automatic-speech-recognition
18
+ - speech
19
+ - audio
20
+ - Transducer
21
+ - TDT
22
+ - FastConformer
23
+ - Conformer
24
+ - pytorch
25
+ - NeMo
26
+ - hf-asr-leaderboard
27
+ license: cc-by-4.0
28
+ widget:
29
+ - example_title: Librispeech sample 1
30
+ src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
31
+ - example_title: Librispeech sample 2
32
+ src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
33
+ model-index:
34
+ - name: parakeet-tdt_ctc-110m
35
+ results:
36
+ - task:
37
+ name: Automatic Speech Recognition
38
+ type: automatic-speech-recognition
39
+ dataset:
40
+ name: AMI (Meetings test)
41
+ type: edinburghcstr/ami
42
+ config: ihm
43
+ split: test
44
+ args:
45
+ language: en
46
+ metrics:
47
+ - name: Test WER
48
+ type: wer
49
+ value: 15.88
50
+ - task:
51
+ name: Automatic Speech Recognition
52
+ type: automatic-speech-recognition
53
+ dataset:
54
+ name: Earnings-22
55
+ type: revdotcom/earnings22
56
+ split: test
57
+ args:
58
+ language: en
59
+ metrics:
60
+ - name: Test WER
61
+ type: wer
62
+ value: 12.42
63
+ - task:
64
+ name: Automatic Speech Recognition
65
+ type: automatic-speech-recognition
66
+ dataset:
67
+ name: GigaSpeech
68
+ type: speechcolab/gigaspeech
69
+ split: test
70
+ args:
71
+ language: en
72
+ metrics:
73
+ - name: Test WER
74
+ type: wer
75
+ value: 10.52
76
+ - task:
77
+ name: Automatic Speech Recognition
78
+ type: automatic-speech-recognition
79
+ dataset:
80
+ name: LibriSpeech (clean)
81
+ type: librispeech_asr
82
+ config: other
83
+ split: test
84
+ args:
85
+ language: en
86
+ metrics:
87
+ - name: Test WER
88
+ type: wer
89
+ value: 2.4
90
+ - task:
91
+ name: Automatic Speech Recognition
92
+ type: automatic-speech-recognition
93
+ dataset:
94
+ name: LibriSpeech (other)
95
+ type: librispeech_asr
96
+ config: other
97
+ split: test
98
+ args:
99
+ language: en
100
+ metrics:
101
+ - name: Test WER
102
+ type: wer
103
+ value: 5.2
104
+ - task:
105
+ type: Automatic Speech Recognition
106
+ name: automatic-speech-recognition
107
+ dataset:
108
+ name: SPGI Speech
109
+ type: kensho/spgispeech
110
+ config: test
111
+ split: test
112
+ args:
113
+ language: en
114
+ metrics:
115
+ - name: Test WER
116
+ type: wer
117
+ value: 2.54
118
+ - task:
119
+ type: Automatic Speech Recognition
120
+ name: automatic-speech-recognition
121
+ dataset:
122
+ name: tedlium-v3
123
+ type: LIUM/tedlium
124
+ config: release1
125
+ split: test
126
+ args:
127
+ language: en
128
+ metrics:
129
+ - name: Test WER
130
+ type: wer
131
+ value: 4.16
132
+ - task:
133
+ name: Automatic Speech Recognition
134
+ type: automatic-speech-recognition
135
+ dataset:
136
+ name: Vox Populi
137
+ type: facebook/voxpopuli
138
+ config: en
139
+ split: test
140
+ args:
141
+ language: en
142
+ metrics:
143
+ - name: Test WER
144
+ type: wer
145
+ value: 6.91
146
+
147
+ metrics:
148
+ - wer
149
+ pipeline_tag: automatic-speech-recognition
150
+ ---
151
+
152
  # Parakeet-TDT-CTC-110M CoreML
153
 
154
  NVIDIA's Parakeet-TDT-CTC-110M model converted to CoreML format for efficient inference on Apple Silicon.