--- library_name: pytorch license: other tags: - foundation - android pipeline_tag: automatic-speech-recognition --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_base_en/web-assets/model_demo.png) # Whisper-Base-En: Optimized for Mobile Deployment ## Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace 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 model is an implementation of Whisper-Base-En found [here](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/whisper). This repository provides scripts to run Whisper-Base-En on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/whisper_base_en). ### Model Details - **Model Type:** Model_use_case.speech_recognition - **Model Stats:** - Model checkpoint: openai/whisper-base - Input resolution: 80x3000 (30 seconds audio) - Max decoded sequence length: 200 tokens - Number of parameters (HfWhisperEncoder): 23.7M - Model size (HfWhisperEncoder) (float): 90.7 MB - Number of parameters (HfWhisperDecoder): 48.9M - Model size (HfWhisperDecoder) (float): 187 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | HfWhisperEncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 863.839 ms | 38 - 66 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 130.073 ms | 0 - 130 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 292.604 ms | 38 - 111 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 112.638 ms | 1 - 130 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 234.299 ms | 0 - 69 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 43.537 ms | 0 - 45 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 400.795 ms | 38 - 67 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 50.197 ms | 1 - 133 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 863.839 ms | 38 - 66 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 130.073 ms | 0 - 130 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 257.222 ms | 0 - 71 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 43.566 ms | 0 - 43 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 229.608 ms | 38 - 74 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 92.405 ms | 0 - 139 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 229.942 ms | 0 - 68 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 43.746 ms | 0 - 42 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 400.795 ms | 38 - 67 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 50.197 ms | 1 - 133 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 230.509 ms | 38 - 114 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 43.232 ms | 1 - 41 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 174.142 ms | 38 - 105 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 34.16 ms | 0 - 135 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 171.96 ms | 37 - 67 MB | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperEncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 27.427 ms | 0 - 129 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperEncoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 42.083 ms | 33 - 33 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 9.663 ms | 5 - 153 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 6.259 ms | 9 - 62 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 7.142 ms | 5 - 195 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 5.165 ms | 17 - 71 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 6.046 ms | 2 - 388 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 3.874 ms | 20 - 32 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 6.999 ms | 5 - 154 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 4.701 ms | 19 - 69 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 9.663 ms | 5 - 153 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 6.259 ms | 9 - 62 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 5.937 ms | 2 - 373 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 4.004 ms | 20 - 33 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 7.372 ms | 5 - 151 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 5.244 ms | 15 - 113 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 5.978 ms | 2 - 328 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 3.905 ms | 20 - 32 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 6.999 ms | 5 - 154 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 4.701 ms | 19 - 69 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 5.929 ms | 2 - 249 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 3.978 ms | 20 - 32 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 4.517 ms | 3 - 190 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 3.178 ms | 5 - 62 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 4.061 ms | 3 - 154 MB | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.tflite) | | HfWhisperDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.622 ms | 20 - 80 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | | HfWhisperDecoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 3.57 ms | 150 - 150 MB | NPU | [Whisper-Base-En.dlc](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/Whisper-Base-En.dlc) | ## Installation Install the package via pip: ```bash pip install "qai-hub-models[whisper-base-en]" ``` ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) 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. ```bash qai-hub configure --api_token API_TOKEN ``` Navigate to [docs](https://app.aihub.qualcomm.com/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. ```bash python -m qai_hub_models.models.whisper_base_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_base_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. ```bash python -m qai_hub_models.models.whisper_base_en.export ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/whisper_base_en/qai_hub_models/models/Whisper-Base-En/export.py) leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) 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. ```python import torch import qai_hub as hub from qai_hub_models.models.whisper_base_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. ```python 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. ```python 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](https://myaccount.qualcomm.com/signup). ## Deploying compiled model to Android The models can be deployed using multiple runtimes: - TensorFlow Lite (`.tflite` export): [This tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a guide to deploy the .tflite model in an Android application. - QNN (`.so` export ): This [sample app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html) provides instructions on how to use the `.so` shared library in an Android application. ## View on Qualcomm® AI Hub Get more details on Whisper-Base-En's performance across various devices [here](https://aihub.qualcomm.com/models/whisper_base_en). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of Whisper-Base-En can be found [here](https://github.com/huggingface/transformers/blob/v4.42.3/LICENSE). * 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) ## References * [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) * [Source Model Implementation](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/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).