Hugging Face's logo

qualcomm
/
Whisper-Tiny

Automatic Speech Recognition
PyTorch
foundation
android
Model card Files
xet
 Community

Whisper-Tiny: Optimized for Qualcomm Devices

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.

This is based on the implementation of Whisper-Tiny found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up 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
PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Snapdragon® X Elite QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 1 Mobile QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Mobile QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.45, ONNX Runtime 1.25.0 Download
PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 QAIRT 2.45, ONNX Runtime 1.25.0 Download
QNN_CONTEXT_BINARY float Snapdragon® X2 Elite QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Snapdragon® X Elite QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 1 Mobile QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Qualcomm® SA8775P QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Mobile QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Qualcomm® SA7255P QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Qualcomm® SA8295P QAIRT 2.45 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS9075 QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit Whisper-Tiny on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models 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 Whisper-Tiny on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.speech_recognition

Model Stats:

  • Model checkpoint: openai/whisper-tiny
  • Input resolution: 80x3000 (30 seconds audio)
  • Max decoded sequence length: 200 tokens
  • Number of parameters (encoder): 9.39M
  • Model size (encoder) (float): 35.9 MB
  • Number of parameters (decoder): 28.4M
  • Model size (decoder) (float): 109 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
decoder PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite 1.419 ms 10 - 10 MB NPU
decoder PRECOMPILED_QNN_ONNX float Snapdragon® X Elite 2.183 ms 84 - 84 MB NPU
decoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile 1.939 ms 0 - 7 MB NPU
decoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 1 Mobile 2.795 ms 13 - 25 MB NPU
decoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) 2.536 ms 10 - 13 MB NPU
decoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8450 2.795 ms 13 - 25 MB NPU
decoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile 1.518 ms 1 - 8 MB NPU
decoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 3.174 ms 10 - 23 MB NPU
decoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Mobile 1.609 ms 0 - 7 MB NPU
decoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8750 1.609 ms 0 - 7 MB NPU
decoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS7181 2.183 ms 84 - 84 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® X2 Elite 1.741 ms 10 - 10 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® X Elite 2.29 ms 10 - 10 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile 1.866 ms 10 - 19 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 1 Mobile 2.608 ms 0 - 11 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS8275 4.593 ms 10 - 18 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) 2.459 ms 10 - 11 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS8450 2.608 ms 0 - 11 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile 1.489 ms 0 - 10 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® SA7255P 4.593 ms 10 - 18 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS9075 3.014 ms 10 - 23 MB NPU
decoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Mobile 1.574 ms 0 - 13 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® SA8295P 2.913 ms 8 - 13 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS8750 1.574 ms 0 - 13 MB NPU
decoder QNN_CONTEXT_BINARY float Qualcomm® QCS7181 2.29 ms 10 - 10 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite 12.608 ms 33 - 33 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® X Elite 26.152 ms 33 - 33 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile 20.025 ms 21 - 28 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 1 Mobile 44.193 ms 20 - 34 MB NPU
encoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) 25.382 ms 12 - 33 MB NPU
encoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8450 44.193 ms 20 - 34 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile 12.497 ms 21 - 28 MB NPU
encoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 29.166 ms 20 - 23 MB NPU
encoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Mobile 15.52 ms 20 - 28 MB NPU
encoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8750 15.52 ms 20 - 28 MB NPU
encoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS7181 26.152 ms 33 - 33 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® X2 Elite 13.118 ms 0 - 0 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® X Elite 26.304 ms 0 - 0 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile 19.99 ms 0 - 8 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 1 Mobile 44.035 ms 1 - 11 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS8275 73.01 ms 1 - 10 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) 25.307 ms 1 - 7 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS8450 44.035 ms 1 - 11 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile 12.482 ms 1 - 10 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® SA7255P 73.01 ms 1 - 10 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS9075 29.142 ms 2 - 13 MB NPU
encoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Mobile 15.426 ms 1 - 14 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® SA8295P 42.408 ms 0 - 6 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS8750 15.426 ms 1 - 14 MB NPU
encoder QNN_CONTEXT_BINARY float Qualcomm® QCS7181 26.304 ms 0 - 0 MB NPU

License

  • The license for the original implementation of Whisper-Tiny can be found here.

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Automatic Speech Recognition
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support