Hugging Face's logo

qualcomm
/
Whisper-Small-En

Automatic Speech Recognition
PyTorch
LiteRT
ONNX
foundation
android
Model card Files Community

Whisper-Small-En: Optimized for Mobile Deployment

Automatic speech recognition (ASR) model for English transcription as well as translation

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.

This model is an implementation of Whisper-Small-En found here.

This repository provides scripts to run Whisper-Small-En on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

  • Model Type: Model_use_case.speech_recognition
  • Model Stats:
    • Model checkpoint: small.en
    • Input resolution: 80x3000 (30 seconds audio)
    • Mean decoded sequence length: 112 tokens
    • Number of parameters (WhisperEncoder): 102M
    • Model size (WhisperEncoder): 390 MB
    • Number of parameters (WhisperDecoder): 139M
    • Model size (WhisperDecoder): 531 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
WhisperEncoderInf float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 3226.414 ms 108 - 141 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 986.224 ms 101 - 200 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 1103.354 ms 74 - 155 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 1271.289 ms 100 - 133 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float SA7255P ADP Qualcomm® SA7255P TFLITE 3226.414 ms 108 - 141 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 779.384 ms 24 - 121 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float SA8295P ADP Qualcomm® SA8295P TFLITE 656.617 ms 109 - 142 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 699.122 ms 110 - 183 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float SA8775P ADP Qualcomm® SA8775P TFLITE 1271.289 ms 100 - 133 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 1245.355 ms 0 - 164 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 525.323 ms 110 - 200 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 542.231 ms 111 - 141 MB GPU Whisper-Small-En.tflite
WhisperEncoderInf float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 609.485 ms 295 - 295 MB NPU Whisper-Small-En.onnx
WhisperDecoderInf float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 65.318 ms 16 - 268 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 62.718 ms 16 - 404 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 55.629 ms 16 - 43 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 52.514 ms 16 - 268 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float SA7255P ADP Qualcomm® SA7255P TFLITE 65.318 ms 16 - 268 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 54.617 ms 16 - 44 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float SA8295P ADP Qualcomm® SA8295P TFLITE 56.316 ms 16 - 248 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 55.232 ms 14 - 40 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float SA8775P ADP Qualcomm® SA8775P TFLITE 52.514 ms 16 - 268 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 55.786 ms 16 - 45 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 45.752 ms 16 - 412 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 42.151 ms 23 - 278 MB NPU Whisper-Small-En.tflite
WhisperDecoderInf float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 48.51 ms 226 - 226 MB NPU Whisper-Small-En.onnx

Installation

Install the package via pip:

pip install "qai-hub-models[whisper-small-en]"

Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to Qualcomm® AI Hub with your Qualcomm® ID. Once signed in navigate to Account -> Settings -> API Token.

With this API token, you can configure your client to run models on the cloud hosted devices.

qai-hub configure --api_token API_TOKEN

Navigate to docs for more information.

Demo off target

The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input.

python -m qai_hub_models.models.whisper_small_en.demo

The above demo runs a reference implementation of pre-processing, model inference, and post processing.

NOTE: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above).

%run -m qai_hub_models.models.whisper_small_en.demo

Run model on a cloud-hosted device

In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following:

  • Performance check on-device on a cloud-hosted device
  • Downloads compiled assets that can be deployed on-device for Android.
  • Accuracy check between PyTorch and on-device outputs.
python -m qai_hub_models.models.whisper_small_en.export

Profiling Results
------------------------------------------------------------
WhisperEncoderInf
Device                          : cs_8275 (ANDROID 14)                  
Runtime                         : TFLITE                                
Estimated inference time (ms)   : 3226.4                                
Estimated peak memory usage (MB): [108, 141]                            
Total # Ops                     : 911                                   
Compute Unit(s)                 : npu (0 ops) gpu (900 ops) cpu (11 ops)

------------------------------------------------------------
WhisperDecoderInf
Device                          : cs_8275 (ANDROID 14)                  
Runtime                         : TFLITE                                
Estimated inference time (ms)   : 65.3                                  
Estimated peak memory usage (MB): [16, 268]                             
Total # Ops                     : 2573                                  
Compute Unit(s)                 : npu (2573 ops) gpu (0 ops) cpu (0 ops)

How does this work?

This export script leverages Qualcomm® AI Hub to optimize, validate, and deploy this model on-device. Lets go through each step below in detail:

Step 1: Compile model for on-device deployment

To compile a PyTorch model for on-device deployment, we first trace the model in memory using the jit.trace and then call the submit_compile_job API.

import torch

import qai_hub as hub
from qai_hub_models.models.whisper_small_en import Model

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S24")

# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()

pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])

# Compile model on a specific device
compile_job = hub.submit_compile_job(
    model=pt_model,
    device=device,
    input_specs=torch_model.get_input_spec(),
)

# Get target model to run on-device
target_model = compile_job.get_target_model()

Step 2: Performance profiling on cloud-hosted device

After compiling models from step 1. Models can be profiled model on-device using the target_model. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics.

profile_job = hub.submit_profile_job(
    model=target_model,
    device=device,
)

Step 3: Verify on-device accuracy

To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device.

input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = inference_job.download_output_data()

With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output.

Note: This on-device profiling and inference requires access to Qualcomm® AI Hub. Sign up for access.

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

  • TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.

  • QNN (.so export ): This sample app provides instructions on how to use the .so shared library in an Android application.

View on Qualcomm® AI Hub

Get more details on Whisper-Small-En's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

  • The license for the original implementation of Whisper-Small-En can be found here.
  • The license for the compiled assets for on-device deployment 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