Raivis Dejus commited on
Commit
d487776
1 Parent(s): 6ddc540
README.md CHANGED
@@ -11,7 +11,7 @@ metrics:
11
  pipeline_tag: automatic-speech-recognition
12
  ---
13
 
14
- # Latvian Whisper tiny speech recognition model
15
 
16
  Trained on combination of:
17
  - Common Voice 17, custom selection of all validated clips, max 1000 clips per speaker
@@ -19,4 +19,4 @@ Trained on combination of:
19
 
20
  Both regular whisper model and CTranslate2 converted version for use with [faster-whisper](https://github.com/SYSTRAN/faster-whisper) as part of [Home Assistant Whisper integration](https://www.home-assistant.io/integrations/whisper/) are available.
21
 
22
- Speech recognition quality is poor, more data is needed, donate your voice on [Balsu talka](https://balsutalka.lv/)
 
11
  pipeline_tag: automatic-speech-recognition
12
  ---
13
 
14
+ # Latvian Whisper small speech recognition model
15
 
16
  Trained on combination of:
17
  - Common Voice 17, custom selection of all validated clips, max 1000 clips per speaker
 
19
 
20
  Both regular whisper model and CTranslate2 converted version for use with [faster-whisper](https://github.com/SYSTRAN/faster-whisper) as part of [Home Assistant Whisper integration](https://www.home-assistant.io/integrations/whisper/) are available.
21
 
22
+ To improve speech recognition quality, more data is needed, donate your voice on [Balsu talka](https://balsutalka.lv/)
config.json CHANGED
@@ -49,7 +49,7 @@
49
  "use_cache": true,
50
  "use_weighted_layer_sum": false,
51
  "vocab_size": 51865,
52
- "alignment_heads": [
53
  [
54
  5,
55
  3
 
49
  "use_cache": true,
50
  "use_weighted_layer_sum": false,
51
  "vocab_size": 51865,
52
+ "alignment_heads": [
53
  [
54
  5,
55
  3
convert-to-safetensors.py → fix-model-metadata.py RENAMED
@@ -1,15 +1,9 @@
1
- # pip install git+https://github.com/openai/whisper.git
2
- # pip install safetensors
3
-
4
- import whisper
5
- import safetensors.torch
6
-
7
- model = whisper.load_model("small")
8
- safetensors.torch.save_model(model, "model.safetensors")
9
 
10
  tensors = dict()
11
  with safetensors.safe_open("./model.safetensors", framework="pt") as f:
12
  for key in f.keys():
13
  tensors[key] = f.get_tensor(key)
14
 
15
- safetensors.torch.save_file(tensors, "./model.safetensors", metadata={'format': 'pt'})
 
1
+ import safetensors
2
+ from safetensors.torch import save_file
 
 
 
 
 
 
3
 
4
  tensors = dict()
5
  with safetensors.safe_open("./model.safetensors", framework="pt") as f:
6
  for key in f.keys():
7
  tensors[key] = f.get_tensor(key)
8
 
9
+ save_file(tensors, "./model.safetensors", metadata={'format': 'pt'})
model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:87b4355275c56e5bdc097320ac7c0131d88058057851a792eae0109da45d26dd
3
- size 482976402
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df9ddf80e8488036fe50bcbabd56affd4e6159223faa4c197aaf3553ccfcb376
3
+ size 483547016
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:60c79be00a6148f08f8147384322114b3fefe51f22d2e39af30ea8eae83dffa7
3
- size 966989264
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:623d0cc8bde553c1a83f7165fdc927c3f03dacd73ee565506d341c84a47b6d7a
3
+ size 966995080