Hugging Face's logo

qualcomm
/
Whisper-Tiny-En

Automatic Speech Recognition
PyTorch
TF Lite
foundation
android
Model card Files Community
qaihm-bot commited on
Commit
e7ec33b
·
verified ·
1 Parent(s): 19fdd6b

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +46 -46
README.md CHANGED
@@ -38,48 +38,48 @@ More details on model performance across various devices, can be found
38
 
39
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
- | WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 4.05 ms | 3 - 40 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
42
- | WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.401 ms | 0 - 121 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.so) |
43
- | WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 3.05 ms | 0 - 62 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
44
- | WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.71 ms | 4 - 49 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.so) |
45
- | WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 2.62 ms | 0 - 56 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
46
- | WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.546 ms | 0 - 42 MB | FP16 | NPU | Use Export Script |
47
- | WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 4.107 ms | 3 - 41 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
48
- | WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 2.368 ms | 10 - 12 MB | FP16 | NPU | Use Export Script |
49
- | WhisperDecoder | SA7255P ADP | SA7255P | TFLITE | 18.307 ms | 2 - 57 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
50
- | WhisperDecoder | SA7255P ADP | SA7255P | QNN | 15.533 ms | 9 - 19 MB | FP16 | NPU | Use Export Script |
51
- | WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 4.007 ms | 3 - 40 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
52
- | WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 2.339 ms | 3 - 4 MB | FP16 | NPU | Use Export Script |
53
- | WhisperDecoder | SA8295P ADP | SA8295P | TFLITE | 5.188 ms | 3 - 55 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
54
- | WhisperDecoder | SA8295P ADP | SA8295P | QNN | 3.586 ms | 1 - 7 MB | FP16 | NPU | Use Export Script |
55
- | WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 4.064 ms | 3 - 41 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
56
- | WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 2.422 ms | 10 - 11 MB | FP16 | NPU | Use Export Script |
57
- | WhisperDecoder | SA8775P ADP | SA8775P | TFLITE | 5.443 ms | 0 - 55 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
58
- | WhisperDecoder | SA8775P ADP | SA8775P | QNN | 3.465 ms | 9 - 15 MB | FP16 | NPU | Use Export Script |
59
- | WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 4.664 ms | 3 - 62 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
60
- | WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.892 ms | 4 - 50 MB | FP16 | NPU | Use Export Script |
61
- | WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 2.285 ms | 10 - 10 MB | FP16 | NPU | Use Export Script |
62
- | WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 99.892 ms | 20 - 51 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
63
- | WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 139.106 ms | 0 - 55 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.so) |
64
- | WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 81.683 ms | 17 - 47 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
65
- | WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 109.268 ms | 0 - 191 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.so) |
66
- | WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 74.784 ms | 25 - 44 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
67
- | WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 98.55 ms | 0 - 195 MB | FP16 | NPU | Use Export Script |
68
- | WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 96.819 ms | 13 - 55 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
69
- | WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 104.143 ms | 0 - 5 MB | FP16 | NPU | Use Export Script |
70
- | WhisperEncoder | SA7255P ADP | SA7255P | TFLITE | 507.645 ms | 20 - 45 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
71
- | WhisperEncoder | SA7255P ADP | SA7255P | QNN | 464.481 ms | 1 - 10 MB | FP16 | NPU | Use Export Script |
72
- | WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 100.562 ms | 18 - 147 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
73
- | WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 106.657 ms | 0 - 5 MB | FP16 | NPU | Use Export Script |
74
- | WhisperEncoder | SA8295P ADP | SA8295P | TFLITE | 103.764 ms | 21 - 42 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
75
- | WhisperEncoder | SA8295P ADP | SA8295P | QNN | 127.685 ms | 4 - 10 MB | FP16 | NPU | Use Export Script |
76
- | WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 109.307 ms | 20 - 60 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
77
- | WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 107.644 ms | 0 - 5 MB | FP16 | NPU | Use Export Script |
78
- | WhisperEncoder | SA8775P ADP | SA8775P | TFLITE | 177.953 ms | 20 - 47 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
79
- | WhisperEncoder | SA8775P ADP | SA8775P | QNN | 119.437 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
80
- | WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 185.481 ms | 11 - 48 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
81
- | WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 165.323 ms | 0 - 196 MB | FP16 | NPU | Use Export Script |
82
- | WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 97.277 ms | 0 - 0 MB | FP16 | NPU | Use Export Script |
83
 
84
 
85
 
@@ -144,8 +144,8 @@ Profiling Results
144
  WhisperDecoder
145
  Device : Samsung Galaxy S23 (13)
146
  Runtime : TFLITE
147
- Estimated inference time (ms) : 4.0
148
- Estimated peak memory usage (MB): [3, 40]
149
  Total # Ops : 557
150
  Compute Unit(s) : NPU (557 ops)
151
 
@@ -153,8 +153,8 @@ Compute Unit(s) : NPU (557 ops)
153
  WhisperEncoder
154
  Device : Samsung Galaxy S23 (13)
155
  Runtime : TFLITE
156
- Estimated inference time (ms) : 99.9
157
- Estimated peak memory usage (MB): [20, 51]
158
  Total # Ops : 271
159
  Compute Unit(s) : GPU (260 ops) CPU (11 ops)
160
  ```
 
38
 
39
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
+ | WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 3.931 ms | 4 - 42 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
42
+ | WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.387 ms | 6 - 76 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.so) |
43
+ | WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 3.056 ms | 0 - 62 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
44
+ | WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.779 ms | 4 - 48 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.so) |
45
+ | WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 2.61 ms | 2 - 59 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
46
+ | WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.425 ms | 0 - 42 MB | FP16 | NPU | Use Export Script |
47
+ | WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 3.987 ms | 3 - 39 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
48
+ | WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 2.337 ms | 10 - 12 MB | FP16 | NPU | Use Export Script |
49
+ | WhisperDecoder | SA7255P ADP | SA7255P | TFLITE | 18.219 ms | 3 - 57 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
50
+ | WhisperDecoder | SA7255P ADP | SA7255P | QNN | 15.609 ms | 9 - 18 MB | FP16 | NPU | Use Export Script |
51
+ | WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 4.113 ms | 3 - 41 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
52
+ | WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 2.499 ms | 10 - 13 MB | FP16 | NPU | Use Export Script |
53
+ | WhisperDecoder | SA8295P ADP | SA8295P | TFLITE | 5.206 ms | 2 - 56 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
54
+ | WhisperDecoder | SA8295P ADP | SA8295P | QNN | 3.37 ms | 3 - 17 MB | FP16 | NPU | Use Export Script |
55
+ | WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 4.048 ms | 3 - 41 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
56
+ | WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 2.393 ms | 10 - 13 MB | FP16 | NPU | Use Export Script |
57
+ | WhisperDecoder | SA8775P ADP | SA8775P | TFLITE | 5.318 ms | 3 - 57 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
58
+ | WhisperDecoder | SA8775P ADP | SA8775P | QNN | 3.536 ms | 9 - 19 MB | FP16 | NPU | Use Export Script |
59
+ | WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 4.755 ms | 3 - 59 MB | FP16 | NPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperDecoder.tflite) |
60
+ | WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.814 ms | 10 - 58 MB | FP16 | NPU | Use Export Script |
61
+ | WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 2.372 ms | 10 - 10 MB | FP16 | NPU | Use Export Script |
62
+ | WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 101.153 ms | 20 - 69 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
63
+ | WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 135.122 ms | 0 - 51 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.so) |
64
+ | WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 82.296 ms | 18 - 55 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
65
+ | WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 108.886 ms | 0 - 191 MB | FP16 | NPU | [Whisper-Tiny-En.so](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.so) |
66
+ | WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 71.468 ms | 20 - 44 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
67
+ | WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 89.444 ms | 0 - 195 MB | FP16 | NPU | Use Export Script |
68
+ | WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 102.293 ms | 14 - 61 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
69
+ | WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 105.262 ms | 0 - 3 MB | FP16 | NPU | Use Export Script |
70
+ | WhisperEncoder | SA7255P ADP | SA7255P | TFLITE | 522.027 ms | 20 - 43 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
71
+ | WhisperEncoder | SA7255P ADP | SA7255P | QNN | 464.37 ms | 1 - 9 MB | FP16 | NPU | Use Export Script |
72
+ | WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 103.025 ms | 20 - 69 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
73
+ | WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 105.663 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
74
+ | WhisperEncoder | SA8295P ADP | SA8295P | TFLITE | 107.129 ms | 21 - 50 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
75
+ | WhisperEncoder | SA8295P ADP | SA8295P | QNN | 127.987 ms | 1 - 15 MB | FP16 | NPU | Use Export Script |
76
+ | WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 100.793 ms | 20 - 69 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
77
+ | WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 106.469 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
78
+ | WhisperEncoder | SA8775P ADP | SA8775P | TFLITE | 179.876 ms | 20 - 42 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
79
+ | WhisperEncoder | SA8775P ADP | SA8775P | QNN | 119.73 ms | 1 - 11 MB | FP16 | NPU | Use Export Script |
80
+ | WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 143.868 ms | 20 - 64 MB | FP16 | GPU | [Whisper-Tiny-En.tflite](https://huggingface.co/qualcomm/Whisper-Tiny-En/blob/main/WhisperEncoder.tflite) |
81
+ | WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 167.53 ms | 0 - 194 MB | FP16 | NPU | Use Export Script |
82
+ | WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 97.829 ms | 0 - 0 MB | FP16 | NPU | Use Export Script |
83
 
84
 
85
 
 
144
  WhisperDecoder
145
  Device : Samsung Galaxy S23 (13)
146
  Runtime : TFLITE
147
+ Estimated inference time (ms) : 3.9
148
+ Estimated peak memory usage (MB): [4, 42]
149
  Total # Ops : 557
150
  Compute Unit(s) : NPU (557 ops)
151
 
 
153
  WhisperEncoder
154
  Device : Samsung Galaxy S23 (13)
155
  Runtime : TFLITE
156
+ Estimated inference time (ms) : 101.2
157
+ Estimated peak memory usage (MB): [20, 69]
158
  Total # Ops : 271
159
  Compute Unit(s) : GPU (260 ops) CPU (11 ops)
160
  ```