Hugging Face's logo

qualcomm
/
Distil-Whisper

Automatic Speech Recognition
PyTorch
android
Model card Files
xet
 Community
Distil-Whisper / README.md
qaihm-bot's picture
qaihm-bot
v0.54.0
3b7bbad verified
preview code
|
raw
history blame contribute delete
10.1 kB
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: automatic-speech-recognition
---
![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/distil_whisper/web-assets/model_demo.png)
# Distil-Whisper: Optimized for Qualcomm Devices
Distil-Whisper Small English is a distilled version of Whisper Small, optimized for fast and efficient automatic speech recognition.
This is based on the implementation of Distil-Whisper found [here](https://github.com/huggingface/distil-whisper).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/distil_whisper) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
## Getting Started
There are two ways to deploy this model on your device:
### Option 1: Download Pre-Exported Models
Below are pre-exported model assets ready for deployment.
| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/distil_whisper/releases/v0.54.0/distil_whisper-onnx-float.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/distil_whisper/releases/v0.54.0/distil_whisper-qnn_dlc-float.zip)
| TFLITE | float | Universal | | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/distil_whisper/releases/v0.54.0/distil_whisper-tflite-float.zip)
For more device-specific assets and performance metrics, visit **[Distil-Whisper on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/distil_whisper)**.
### Option 2: Export with Custom Configurations
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/distil_whisper) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations
This option is ideal if you need to customize the model beyond the default configuration provided here.
See our repository for [Distil-Whisper on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/distil_whisper) for usage instructions.
## Model Details
**Model Type:** Model_use_case.speech_recognition
**Model Stats:**
- Model checkpoint: distil-whisper/distil-small.en
- Input resolution: 80x3000 (30 seconds audio)
- Max decoded sequence length: 200 tokens
- Number of parameters (encoder): 166M
- Model size (encoder) (float): 332 MB
- Number of parameters (decoder): 211M
- Model size (decoder) (float): 450MB
## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| decoder | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.546 ms | 52 - 375 MB | NPU
| decoder | ONNX | float | Snapdragon® 8 Elite Mobile | 7.178 ms | 16 - 472 MB | NPU
| decoder | ONNX | float | Snapdragon® X2 Elite | 5.083 ms | 178 - 178 MB | NPU
| decoder | ONNX | float | Snapdragon® X Elite | 11.41 ms | 178 - 178 MB | NPU
| decoder | ONNX | float | Snapdragon® X Elite | 11.41 ms | 178 - 178 MB | NPU
| decoder | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.67 ms | 52 - 426 MB | NPU
| decoder | ONNX | float | Qualcomm® QCS8550 (Proxy) | 11.746 ms | 0 - 184 MB | NPU
| decoder | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 7.178 ms | 16 - 472 MB | NPU
| decoder | ONNX | float | Qualcomm® QCS9075 | 13.188 ms | 40 - 82 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.934 ms | 39 - 538 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 7.351 ms | 4 - 548 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® X2 Elite | 5.954 ms | 40 - 40 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® X Elite | 10.825 ms | 40 - 40 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® X Elite | 10.825 ms | 40 - 40 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 8.656 ms | 0 - 600 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 11.398 ms | 40 - 634 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® SA8775P | 12.874 ms | 15 - 507 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® SA8775P | 12.874 ms | 15 - 507 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® SA8775P | 12.874 ms | 15 - 507 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® SA7255P | 19.195 ms | 31 - 528 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® SA8295P | 13.997 ms | 21 - 263 MB | NPU
| decoder | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 7.351 ms | 4 - 548 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® QCS9075 | 16.524 ms | 40 - 86 MB | NPU
| decoder | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 18.145 ms | 36 - 339 MB | NPU
| decoder | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.848 ms | 4 - 570 MB | NPU
| decoder | TFLITE | float | Snapdragon® 8 Elite Mobile | 7.215 ms | 4 - 572 MB | NPU
| decoder | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 8.619 ms | 4 - 749 MB | NPU
| decoder | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 11.713 ms | 5 - 7 MB | NPU
| decoder | TFLITE | float | Qualcomm® SA8775P | 13.037 ms | 5 - 537 MB | NPU
| decoder | TFLITE | float | Qualcomm® SA8775P | 13.037 ms | 5 - 537 MB | NPU
| decoder | TFLITE | float | Qualcomm® SA8775P | 13.037 ms | 5 - 537 MB | NPU
| decoder | TFLITE | float | Qualcomm® SA7255P | 19.275 ms | 4 - 537 MB | NPU
| decoder | TFLITE | float | Qualcomm® SA8295P | 14.231 ms | 5 - 297 MB | NPU
| decoder | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 7.215 ms | 4 - 572 MB | NPU
| decoder | TFLITE | float | Qualcomm® QCS9075 | 16.188 ms | 0 - 265 MB | NPU
| decoder | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 18.522 ms | 5 - 466 MB | NPU
| encoder | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 49.731 ms | 80 - 839 MB | NPU
| encoder | ONNX | float | Snapdragon® 8 Elite Mobile | 60.575 ms | 80 - 734 MB | NPU
| encoder | ONNX | float | Snapdragon® X2 Elite | 50.437 ms | 183 - 183 MB | NPU
| encoder | ONNX | float | Snapdragon® X Elite | 123.445 ms | 182 - 182 MB | NPU
| encoder | ONNX | float | Snapdragon® X Elite | 123.445 ms | 182 - 182 MB | NPU
| encoder | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 82.214 ms | 80 - 1239 MB | NPU
| encoder | ONNX | float | Qualcomm® QCS8550 (Proxy) | 118.816 ms | 0 - 202 MB | NPU
| encoder | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 60.575 ms | 80 - 734 MB | NPU
| encoder | ONNX | float | Qualcomm® QCS9075 | 150.604 ms | 79 - 83 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 57.546 ms | 1 - 712 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 71.556 ms | 1 - 691 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® X2 Elite | 60.179 ms | 1 - 1 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® X Elite | 139.254 ms | 1 - 1 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® X Elite | 139.254 ms | 1 - 1 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 96.947 ms | 0 - 963 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 135.243 ms | 1 - 4 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® SA8775P | 153.197 ms | 1 - 687 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® SA8775P | 153.197 ms | 1 - 687 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® SA8775P | 153.197 ms | 1 - 687 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® SA7255P | 437.44 ms | 1 - 696 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® SA8295P | 192.943 ms | 1 - 610 MB | NPU
| encoder | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 71.556 ms | 1 - 691 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® QCS9075 | 170.231 ms | 1 - 39 MB | NPU
| encoder | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 267.063 ms | 5 - 828 MB | NPU
| encoder | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 403.997 ms | 42 - 85 MB | GPU
| encoder | TFLITE | float | Snapdragon® 8 Elite Mobile | 409.903 ms | 40 - 80 MB | GPU
| encoder | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 475.908 ms | 42 - 184 MB | GPU
| encoder | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 657.568 ms | 0 - 318 MB | GPU
| encoder | TFLITE | float | Qualcomm® SA8775P | 1316.653 ms | 20 - 64 MB | GPU
| encoder | TFLITE | float | Qualcomm® SA8775P | 1316.653 ms | 20 - 64 MB | GPU
| encoder | TFLITE | float | Qualcomm® SA8775P | 1316.653 ms | 20 - 64 MB | GPU
| encoder | TFLITE | float | Qualcomm® SA7255P | 3135.306 ms | 24 - 69 MB | GPU
| encoder | TFLITE | float | Qualcomm® SA8295P | 671.062 ms | 38 - 81 MB | GPU
| encoder | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 409.903 ms | 40 - 80 MB | GPU
| encoder | TFLITE | float | Qualcomm® QCS9075 | 1271.896 ms | 0 - 40 MB | GPU
| encoder | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 852.126 ms | 39 - 193 MB | GPU
## License
* The license for the original implementation of Distil-Whisper can be found
 [here](https://github.com/huggingface/distil-whisper/blob/main/LICENSE).
## References
* [Distil-Whisper - Robust Knowledge Distillation via Large-Scale Pseudo Labelling](https://arxiv.org/abs/2311.00430)
* [Source Model Implementation](https://github.com/huggingface/distil-whisper)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).