Automatic Speech Recognition
WhisperKit
Core ML
Arabic
Egyptian Arabic
English
whisper
code-switching
Instructions to use zayednetad/whisper-small-ar-en-codeswitching-coreml with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- WhisperKit
How to use zayednetad/whisper-small-ar-en-codeswitching-coreml with WhisperKit:
# Install CLI with Homebrew on macOS device brew install whisperkit-cli # View all available inference options whisperkit-cli transcribe --help # Download and run inference using whisper base model whisperkit-cli transcribe --audio-path /path/to/audio.mp3 # Or use your preferred model variant whisperkit-cli transcribe --model "large-v3" --model-prefix "distil" --audio-path /path/to/audio.mp3 --verbose
- Notebooks
- Google Colab
- Kaggle
Replace with whisperkittools conversion of IbrahimAmin/code-switched-egyptian-arabic-whisper-small (azeem23 conversion failed on-device validation)
Browse files- AudioEncoder.mlmodelc/analytics/coremldata.bin +1 -1
- AudioEncoder.mlmodelc/coremldata.bin +1 -1
- AudioEncoder.mlmodelc/weights/weight.bin +1 -1
- MelSpectrogram.mlmodelc/analytics/coremldata.bin +1 -1
- MelSpectrogram.mlmodelc/coremldata.bin +1 -1
- README.md +10 -5
- TextDecoder.mlmodelc/analytics/coremldata.bin +1 -1
- TextDecoder.mlmodelc/coremldata.bin +1 -1
- TextDecoder.mlmodelc/weights/weight.bin +1 -1
- config.json +1 -1
- generation_config.json +3 -13
AudioEncoder.mlmodelc/analytics/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 243
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68dd2d4061641ea4cd758e8be80fbfa462b92630f866e65e6c463741d0237758
|
| 3 |
size 243
|
AudioEncoder.mlmodelc/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 408
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9955d355cf32b431e988cc09d554e5cc73ada12e2be0161963bfe190166fb8c0
|
| 3 |
size 408
|
AudioEncoder.mlmodelc/weights/weight.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 176323456
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f9d172a8def602b08373af1aca8373ee6b8c7d61614901bff24d3f3a0a1a6538
|
| 3 |
size 176323456
|
MelSpectrogram.mlmodelc/analytics/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 243
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ebae554d74f221aed2c08da402e593fdbb577630f8b258919c5449cc8be25e4
|
| 3 |
size 243
|
MelSpectrogram.mlmodelc/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 389
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f5eed03308eb21ea55d55fe854c4018a7697fb45f6fb93bffe94fd431a0526f
|
| 3 |
size 389
|
README.md
CHANGED
|
@@ -1,9 +1,10 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
| 3 |
language:
|
| 4 |
- ar
|
|
|
|
| 5 |
- en
|
| 6 |
-
base_model:
|
| 7 |
pipeline_tag: automatic-speech-recognition
|
| 8 |
tags:
|
| 9 |
- whisper
|
|
@@ -14,10 +15,14 @@ tags:
|
|
| 14 |
|
| 15 |
# Whisper Small Arabic-English Code-Switching — Core ML
|
| 16 |
|
| 17 |
-
Core ML conversion of [
|
| 18 |
-
(
|
| 19 |
-
|
| 20 |
[whisperkittools](https://github.com/argmaxinc/whisperkittools) for on-device use with
|
| 21 |
[WhisperKit](https://github.com/argmaxinc/WhisperKit).
|
| 22 |
|
| 23 |
Use the standard `openai/whisper-small` tokenizer.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
language:
|
| 4 |
- ar
|
| 5 |
+
- arz
|
| 6 |
- en
|
| 7 |
+
base_model: IbrahimAmin/code-switched-egyptian-arabic-whisper-small
|
| 8 |
pipeline_tag: automatic-speech-recognition
|
| 9 |
tags:
|
| 10 |
- whisper
|
|
|
|
| 15 |
|
| 16 |
# Whisper Small Arabic-English Code-Switching — Core ML
|
| 17 |
|
| 18 |
+
Core ML conversion of [IbrahimAmin/code-switched-egyptian-arabic-whisper-small](https://huggingface.co/IbrahimAmin/code-switched-egyptian-arabic-whisper-small)
|
| 19 |
+
(Apache-2.0, a fine-tune of [openai/whisper-small](https://huggingface.co/openai/whisper-small) for
|
| 20 |
+
Egyptian Arabic-English code-switching), converted with
|
| 21 |
[whisperkittools](https://github.com/argmaxinc/whisperkittools) for on-device use with
|
| 22 |
[WhisperKit](https://github.com/argmaxinc/WhisperKit).
|
| 23 |
|
| 24 |
Use the standard `openai/whisper-small` tokenizer.
|
| 25 |
+
|
| 26 |
+
Earlier commits of this repo contained a conversion of
|
| 27 |
+
azeem23/whisper-small-codeswitching-ArabicEnglish (MIT), replaced after
|
| 28 |
+
on-device validation.
|
TextDecoder.mlmodelc/analytics/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 243
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0139b366f2f968b90262450011800fb6a0cb59dc503a50677d06f366cfeb9050
|
| 3 |
size 243
|
TextDecoder.mlmodelc/coremldata.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 694
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc2e89dc76ce671fdd37235d3f36165a6c99728c9cee20bbfb7b41764b030751
|
| 3 |
size 694
|
TextDecoder.mlmodelc/weights/weight.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 307287346
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec9681d5fafdef76710ab57e8ddaab7b618d7005782836c3d2cbe7ec4cd5dcc5
|
| 3 |
size 307287346
|
config.json
CHANGED
|
@@ -53,7 +53,7 @@
|
|
| 53 |
"pad_token_id": 50257,
|
| 54 |
"scale_embedding": false,
|
| 55 |
"torch_dtype": "float32",
|
| 56 |
-
"transformers_version": "4.
|
| 57 |
"use_cache": true,
|
| 58 |
"use_weighted_layer_sum": false,
|
| 59 |
"vocab_size": 51865
|
|
|
|
| 53 |
"pad_token_id": 50257,
|
| 54 |
"scale_embedding": false,
|
| 55 |
"torch_dtype": "float32",
|
| 56 |
+
"transformers_version": "4.53.0",
|
| 57 |
"use_cache": true,
|
| 58 |
"use_weighted_layer_sum": false,
|
| 59 |
"vocab_size": 51865
|
generation_config.json
CHANGED
|
@@ -48,16 +48,6 @@
|
|
| 48 |
"bos_token_id": 50257,
|
| 49 |
"decoder_start_token_id": 50258,
|
| 50 |
"eos_token_id": 50257,
|
| 51 |
-
"forced_decoder_ids": [
|
| 52 |
-
[
|
| 53 |
-
1,
|
| 54 |
-
null
|
| 55 |
-
],
|
| 56 |
-
[
|
| 57 |
-
2,
|
| 58 |
-
50359
|
| 59 |
-
]
|
| 60 |
-
],
|
| 61 |
"is_multilingual": true,
|
| 62 |
"lang_to_id": {
|
| 63 |
"<|af|>": 50327,
|
|
@@ -160,6 +150,7 @@
|
|
| 160 |
"<|yo|>": 50325,
|
| 161 |
"<|zh|>": 50260
|
| 162 |
},
|
|
|
|
| 163 |
"max_initial_timestamp_index": 50,
|
| 164 |
"max_length": 448,
|
| 165 |
"no_timestamps_token_id": 50363,
|
|
@@ -250,15 +241,14 @@
|
|
| 250 |
49870,
|
| 251 |
50254,
|
| 252 |
50258,
|
| 253 |
-
50358,
|
| 254 |
-
50359,
|
| 255 |
50360,
|
| 256 |
50361,
|
| 257 |
50362
|
| 258 |
],
|
|
|
|
| 259 |
"task_to_id": {
|
| 260 |
"transcribe": 50359,
|
| 261 |
"translate": 50358
|
| 262 |
},
|
| 263 |
-
"transformers_version": "4.
|
| 264 |
}
|
|
|
|
| 48 |
"bos_token_id": 50257,
|
| 49 |
"decoder_start_token_id": 50258,
|
| 50 |
"eos_token_id": 50257,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
"is_multilingual": true,
|
| 52 |
"lang_to_id": {
|
| 53 |
"<|af|>": 50327,
|
|
|
|
| 150 |
"<|yo|>": 50325,
|
| 151 |
"<|zh|>": 50260
|
| 152 |
},
|
| 153 |
+
"language": "arabic",
|
| 154 |
"max_initial_timestamp_index": 50,
|
| 155 |
"max_length": 448,
|
| 156 |
"no_timestamps_token_id": 50363,
|
|
|
|
| 241 |
49870,
|
| 242 |
50254,
|
| 243 |
50258,
|
|
|
|
|
|
|
| 244 |
50360,
|
| 245 |
50361,
|
| 246 |
50362
|
| 247 |
],
|
| 248 |
+
"task": "transcribe",
|
| 249 |
"task_to_id": {
|
| 250 |
"transcribe": 50359,
|
| 251 |
"translate": 50358
|
| 252 |
},
|
| 253 |
+
"transformers_version": "4.53.0"
|
| 254 |
}
|