Hugging Face's logo

qualcomm
/
Whisper-Small-En

Automatic Speech Recognition
PyTorch
LiteRT
foundation
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
d9c6f2e
·
verified ·
1 Parent(s): 6ea879e

v0.35.0

Browse files

See https://github.com/quic/ai-hub-models/releases/v0.35.0 for changelog.

Files changed (13) hide show
  1. .gitattributes +2 -0
  2. LICENSE +1 -1
  3. README.md +57 -33
  4. Whisper-Small-En_WhisperDecoderInf.tflite → Whisper-Small-En_HfWhisperDecoder.dlc +2 -2
  5. Whisper-Small-En_WhisperEncoderInf.tflite → Whisper-Small-En_HfWhisperDecoder.tflite +2 -2
  6. Whisper-Small-En_HfWhisperEncoder.dlc +3 -0
  7. Whisper-Small-En_HfWhisperEncoder.tflite +3 -0
  8. precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperDecoder.bin +3 -0
  9. precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperDecoder.onnx.zip +3 -0
  10. precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperEncoder.bin +3 -0
  11. precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperEncoder.onnx.zip +3 -0
  12. precompiled/qualcomm-snapdragon-x-elite/sdk_versions.yml +5 -0
  13. sdk_versions.yml +5 -0
.gitattributes CHANGED
@@ -38,3 +38,5 @@ WhisperDecoder.so filter=lfs diff=lfs merge=lfs -text
38
  Whisper-Small-En_WhisperDecoderInf.dlc filter=lfs diff=lfs merge=lfs -text
39
  Whisper-Small-En_WhisperEncoderInf.dlc filter=lfs diff=lfs merge=lfs -text
40
  DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text
 
 
 
38
  Whisper-Small-En_WhisperDecoderInf.dlc filter=lfs diff=lfs merge=lfs -text
39
  Whisper-Small-En_WhisperEncoderInf.dlc filter=lfs diff=lfs merge=lfs -text
40
  DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text
41
+ Whisper-Small-En_HfWhisperDecoder.dlc filter=lfs diff=lfs merge=lfs -text
42
+ Whisper-Small-En_HfWhisperEncoder.dlc filter=lfs diff=lfs merge=lfs -text
LICENSE CHANGED
@@ -1,2 +1,2 @@
1
- The license of the original trained model can be found at https://github.com/openai/whisper/blob/main/LICENSE.
2
  The license for the deployable model files (.tflite, .onnx, .dlc, .bin, etc.) can be found in DEPLOYMENT_MODEL_LICENSE.pdf.
 
1
+ The license of the original trained model can be found at https://github.com/huggingface/transformers/blob/v4.42.3/LICENSE.
2
  The license for the deployable model files (.tflite, .onnx, .dlc, .bin, etc.) can be found in DEPLOYMENT_MODEL_LICENSE.pdf.
README.md CHANGED
@@ -11,12 +11,12 @@ pipeline_tag: automatic-speech-recognition
11
  ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_en/web-assets/model_demo.png)
12
 
13
  # Whisper-Small-En: Optimized for Mobile Deployment
14
- ## Automatic speech recognition (ASR) model for English transcription as well as translation
15
 
16
 
17
- OpenAI’s Whisper ASR (Automatic Speech Recognition) model is a state-of-the-art system designed for transcribing spoken language into written text. It exhibits robust performance in realistic, noisy environments, making it highly reliable for real-world applications. Specifically, it excels in long-form transcription, capable of accurately transcribing audio clips up to 30 seconds long. Time to the first token is the encoder's latency, while time to each additional token is decoder's latency, where we assume a mean decoded length specified below.
18
 
19
- This model is an implementation of Whisper-Small-En found [here](https://github.com/openai/whisper/tree/main).
20
 
21
 
22
  This repository provides scripts to run Whisper-Small-En on Qualcomm® devices.
@@ -29,38 +29,62 @@ More details on model performance across various devices, can be found
29
 
30
  - **Model Type:** Model_use_case.speech_recognition
31
  - **Model Stats:**
32
- - Model checkpoint: small.en
33
  - Input resolution: 80x3000 (30 seconds audio)
34
- - Mean decoded sequence length: 112 tokens
35
- - Number of parameters (WhisperEncoderInf): 102M
36
- - Model size (WhisperEncoderInf) (float): 390 MB
37
- - Number of parameters (WhisperDecoderInf): 139M
38
- - Model size (WhisperDecoderInf) (float): 532 MB
39
 
40
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
41
  |---|---|---|---|---|---|---|---|---|
42
- | WhisperEncoderInf | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 3232.589 ms | 109 - 142 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
43
- | WhisperEncoderInf | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 907.681 ms | 110 - 207 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
44
- | WhisperEncoderInf | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 1270.067 ms | 93 - 125 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
45
- | WhisperEncoderInf | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 3232.589 ms | 109 - 142 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
46
- | WhisperEncoderInf | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1123.233 ms | 22 - 77 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
47
- | WhisperEncoderInf | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 656.611 ms | 109 - 142 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
48
- | WhisperEncoderInf | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 685.947 ms | 97 - 119 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
49
- | WhisperEncoderInf | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 1270.067 ms | 93 - 125 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
50
- | WhisperEncoderInf | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 696.19 ms | 65 - 182 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
51
- | WhisperEncoderInf | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 531.086 ms | 108 - 200 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
52
- | WhisperEncoderInf | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 463.534 ms | 109 - 139 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
53
- | WhisperDecoderInf | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 71.04 ms | 16 - 384 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
54
- | WhisperDecoderInf | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 47.2 ms | 16 - 397 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
55
- | WhisperDecoderInf | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 49.446 ms | 16 - 384 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
56
- | WhisperDecoderInf | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 71.04 ms | 16 - 384 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
57
- | WhisperDecoderInf | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 49.535 ms | 16 - 48 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
58
- | WhisperDecoderInf | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 42.734 ms | 16 - 353 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
59
- | WhisperDecoderInf | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 49.183 ms | 11 - 43 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
60
- | WhisperDecoderInf | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 49.446 ms | 16 - 384 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
61
- | WhisperDecoderInf | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 49.082 ms | 8 - 43 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
62
- | WhisperDecoderInf | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 43.076 ms | 16 - 425 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
63
- | WhisperDecoderInf | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 36.925 ms | 15 - 386 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
64
 
65
 
66
 
@@ -221,14 +245,14 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
221
 
222
  ## License
223
  * The license for the original implementation of Whisper-Small-En can be found
224
- [here](https://github.com/openai/whisper/blob/main/LICENSE).
225
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
226
 
227
 
228
 
229
  ## References
230
  * [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf)
231
- * [Source Model Implementation](https://github.com/openai/whisper/tree/main)
232
 
233
 
234
 
 
11
  ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_en/web-assets/model_demo.png)
12
 
13
  # Whisper-Small-En: Optimized for Mobile Deployment
14
+ ## Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace
15
 
16
 
17
+ HuggingFace Whisper-Small ASR (Automatic Speech Recognition) model is a state-of-the-art system designed for transcribing spoken language into written text. This model is based on the transformer architecture and has been optimized for edge inference by replacing Multi-Head Attention (MHA) with Single-Head Attention (SHA) and linear layers with convolutional (conv) layers. It exhibits robust performance in realistic, noisy environments, making it highly reliable for real-world applications. Specifically, it excels in long-form transcription, capable of accurately transcribing audio clips up to 30 seconds long. Time to the first token is the encoder's latency, while time to each additional token is decoder's latency, where we assume a max decoded length specified below.
18
 
19
+ This model is an implementation of Whisper-Small-En found [here](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/whisper).
20
 
21
 
22
  This repository provides scripts to run Whisper-Small-En on Qualcomm® devices.
 
29
 
30
  - **Model Type:** Model_use_case.speech_recognition
31
  - **Model Stats:**
32
+ - Model checkpoint: openai/whisper-small
33
  - Input resolution: 80x3000 (30 seconds audio)
34
+ - Max decoded sequence length: 200 tokens
35
+ - Number of parameters (HfWhisperEncoder): 102M
36
+ - Model size (HfWhisperEncoder) (float): 391 MB
37
+ - Number of parameters (HfWhisperDecoder): 139M
38
+ - Model size (HfWhisperDecoder) (float): 533 MB
39
 
40
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
41
  |---|---|---|---|---|---|---|---|---|
42
+ | HfWhisperEncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 3333.928 ms | 109 - 159 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
43
+ | HfWhisperEncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 424.425 ms | 0 - 443 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
44
+ | HfWhisperEncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1115.67 ms | 18 - 223 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
45
+ | HfWhisperEncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 135.465 ms | 1 - 28 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
46
+ | HfWhisperEncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 1416.508 ms | 106 - 155 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
47
+ | HfWhisperEncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 154.567 ms | 0 - 442 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
48
+ | HfWhisperEncoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 3333.928 ms | 109 - 159 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
49
+ | HfWhisperEncoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 424.425 ms | 0 - 443 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
50
+ | HfWhisperEncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 882.114 ms | 18 - 186 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
51
+ | HfWhisperEncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 137.649 ms | 0 - 33 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
52
+ | HfWhisperEncoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 798.622 ms | 109 - 158 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
53
+ | HfWhisperEncoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 241.302 ms | 0 - 440 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
54
+ | HfWhisperEncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 768.865 ms | 18 - 222 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
55
+ | HfWhisperEncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 138.695 ms | 0 - 32 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
56
+ | HfWhisperEncoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 1416.508 ms | 106 - 155 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
57
+ | HfWhisperEncoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 154.567 ms | 0 - 442 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
58
+ | HfWhisperEncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 775.503 ms | 110 - 130 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
59
+ | HfWhisperEncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 137.037 ms | 0 - 30 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
60
+ | HfWhisperEncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 569.222 ms | 111 - 300 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
61
+ | HfWhisperEncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 106.478 ms | 1 - 449 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
62
+ | HfWhisperEncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 622.531 ms | 110 - 156 MB | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
63
+ | HfWhisperEncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 87.001 ms | 0 - 426 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
64
+ | HfWhisperEncoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 130.938 ms | 158 - 158 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
65
+ | HfWhisperDecoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 28.37 ms | 14 - 506 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
66
+ | HfWhisperDecoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 18.043 ms | 38 - 335 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
67
+ | HfWhisperDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 19.146 ms | 14 - 59 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
68
+ | HfWhisperDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 12.048 ms | 60 - 74 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
69
+ | HfWhisperDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 21.142 ms | 14 - 505 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
70
+ | HfWhisperDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 13.304 ms | 55 - 346 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
71
+ | HfWhisperDecoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 28.37 ms | 14 - 506 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
72
+ | HfWhisperDecoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 18.043 ms | 38 - 335 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
73
+ | HfWhisperDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 19.112 ms | 14 - 40 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
74
+ | HfWhisperDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 11.743 ms | 57 - 83 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
75
+ | HfWhisperDecoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 20.478 ms | 14 - 429 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
76
+ | HfWhisperDecoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 14.571 ms | 47 - 328 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
77
+ | HfWhisperDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 19.092 ms | 0 - 57 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
78
+ | HfWhisperDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.972 ms | 43 - 67 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
79
+ | HfWhisperDecoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 21.142 ms | 14 - 505 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
80
+ | HfWhisperDecoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 13.304 ms | 55 - 346 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
81
+ | HfWhisperDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 18.961 ms | 14 - 66 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
82
+ | HfWhisperDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 11.942 ms | 60 - 85 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
83
+ | HfWhisperDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 15.165 ms | 14 - 668 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
84
+ | HfWhisperDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 9.579 ms | 0 - 308 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
85
+ | HfWhisperDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 13.094 ms | 14 - 459 MB | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.tflite) |
86
+ | HfWhisperDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 8.145 ms | 58 - 169 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
87
+ | HfWhisperDecoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 10.743 ms | 392 - 392 MB | NPU | [Whisper-Small-En.dlc](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/Whisper-Small-En.dlc) |
88
 
89
 
90
 
 
245
 
246
  ## License
247
  * The license for the original implementation of Whisper-Small-En can be found
248
+ [here](https://github.com/huggingface/transformers/blob/v4.42.3/LICENSE).
249
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
250
 
251
 
252
 
253
  ## References
254
  * [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf)
255
+ * [Source Model Implementation](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/whisper)
256
 
257
 
258
 
Whisper-Small-En_WhisperDecoderInf.tflite → Whisper-Small-En_HfWhisperDecoder.dlc RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2c2049a24dbe4c1fcdc354b12d15e1bfad705f70316febd095f8f67c81745fe8
3
- size 557615568
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:812c45f3cb9f19ea44374790a74cb4b06a23f365245c45ace37dd39e5a640649
3
+ size 360926188
Whisper-Small-En_WhisperEncoderInf.tflite → Whisper-Small-En_HfWhisperDecoder.tflite RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2d63084276fa4f2d937797dc9447ef67c8f18814b04848171dac47ad54eeab3a
3
- size 409468768
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a564dddff1d6a0baff6fea614f00947c9d04b8899a3a618c43e45a34aea06ef7
3
+ size 558499300
Whisper-Small-En_HfWhisperEncoder.dlc ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab0c124c09a95e4fc5d6557cceef8228df5155ed5b91b5deb89196c7ced425a1
3
+ size 206618308
Whisper-Small-En_HfWhisperEncoder.tflite ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b27459b4f70e0313de3d3ce61b73677a8629d54e5c2ab52c8ca7a7e17aee8a54
3
+ size 409705372
precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperDecoder.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bb7c14ab5f39b9cab50c29d5deb05f57fa130e52544fd18c9338e306707c566
3
+ size 361385984
precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperDecoder.onnx.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb9c759e32f49891bae61791020c04c2a9d5c23a7c06c1ae3bc25793b10e1bd4
3
+ size 330473992
precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperEncoder.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d4e53de4d1405ec58ade84b8edcd6dd5005394c6d7a78a7fb9c740e38fe92afc
3
+ size 236400640
precompiled/qualcomm-snapdragon-x-elite/Whisper-Small-En_HfWhisperEncoder.onnx.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:397f48562b589fecf3863dc9cf1014430fa58cd9d596f889751e205873da4fa2
3
+ size 197739694
precompiled/qualcomm-snapdragon-x-elite/sdk_versions.yml ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ sdk_versions:
2
+ qnn_context_binary:
3
+ qairt: 2.37.0.250724175447_124859
4
+ precompiled_qnn_onnx:
5
+ qairt: 2.33.2.250410134701_117956
sdk_versions.yml ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ sdk_versions:
2
+ tflite:
3
+ tflite: 2.17.0
4
+ qnn_dlc:
5
+ qairt: 2.37.0.250724175447_124859