Hugging Face's logo

qualcomm
/
Zipformer

Automatic Speech Recognition
PyTorch
foundation
android
Model card Files
xet
Community

Zipformer: Optimized for Qualcomm Devices

Zipformer streaming ASR (Automatic Speech Recognition) model is a state-of-the-art system designed for transcribing spoken language into written text streamingly. This model is based on the transformer architecture and has been optimized for edge inference by replacing 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 audios. Time to the first token is the encoder's latency, while time to each additional token is joiner's latency, where we assume a max decoded length specified below.

This is based on the implementation of Zipformer 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® X Elite QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X Elite QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Gen 3 Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS8550 (Proxy) QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X2 Elite QAIRT 2.42, ONNX Runtime 1.24.1 Download
PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS9075 QAIRT 2.42, ONNX Runtime 1.24.1 Download
QNN_CONTEXT_BINARY float Snapdragon® X Elite QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS8275 (Proxy) QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® SA8775P QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® SA7255P QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® SA8295P QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS9075 QAIRT 2.43 Download
QNN_CONTEXT_BINARY float Qualcomm® QCS8450 (Proxy) QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Snapdragon® X Elite QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Gen 3 Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8275 (Proxy) QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8550 (Proxy) QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Qualcomm® SA8775P QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite Gen 5 Mobile QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Snapdragon® X2 Elite QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Qualcomm® SA7255P QAIRT 2.43 Download
QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS9075 QAIRT 2.43 Download

For more device-specific assets and performance metrics, visit Zipformer 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 Zipformer on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.speech_recognition

Model Stats:

  • Model checkpoint: pfluo/k2fsa-zipformer-chinese-english-mixed
  • Input resolution: 80x71 (0.71 seconds audio)
  • Max decoded sequence length: 200 tokens
  • Number of parameters (ZipformerEncoder): 63.2M
  • Model size (ZipformerEncoder) (float): 242 MB
  • Number of parameters (ZipformerDecoder): 3.47M
  • Model size (ZipformerDecoder) (float): 13.2 MB
  • Number of parameters (ZipformerJoiner): 3.21M
  • Model size (ZipformerJoiner) (float): 12.2 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
ZipformerDecoder PRECOMPILED_QNN_ONNX float Snapdragon® X Elite 0.16 ms 7 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile 0.059 ms 0 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) 0.067 ms 0 - 8 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 0.137 ms 0 - 3 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite For Galaxy Mobile 0.051 ms 0 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile 0.053 ms 0 - 10 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite 0.042 ms 7 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X Elite 0.161 ms 7 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Gen 3 Mobile 0.067 ms 0 - 6 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS8550 (Proxy) 0.078 ms 0 - 1 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS9075 0.19 ms 0 - 3 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite For Galaxy Mobile 0.05 ms 0 - 7 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite Gen 5 Mobile 0.05 ms 0 - 10 MB NPU
ZipformerDecoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X2 Elite 0.039 ms 7 - 7 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Snapdragon® X Elite 0.151 ms 0 - 0 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile 0.063 ms 0 - 7 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® QCS8275 (Proxy) 0.319 ms 0 - 8 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) 0.077 ms 0 - 1 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® SA8775P 0.208 ms 0 - 5 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® QCS9075 0.12 ms 2 - 4 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® QCS8450 (Proxy) 0.102 ms 0 - 9 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® SA7255P 0.319 ms 0 - 8 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Qualcomm® SA8295P 0.33 ms 0 - 5 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite For Galaxy Mobile 0.047 ms 0 - 13 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile 0.041 ms 0 - 10 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Snapdragon® X Elite 0.171 ms 0 - 0 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Gen 3 Mobile 0.056 ms 0 - 7 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8275 (Proxy) 0.322 ms 0 - 8 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8550 (Proxy) 0.063 ms 0 - 1 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Qualcomm® SA8775P 0.215 ms 0 - 5 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS9075 0.134 ms 0 - 2 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Qualcomm® SA7255P 0.322 ms 0 - 8 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite For Galaxy Mobile 0.046 ms 0 - 9 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite Gen 5 Mobile 0.053 ms 0 - 10 MB NPU
ZipformerDecoder QNN_CONTEXT_BINARY w16a16 Snapdragon® X2 Elite 0.144 ms 0 - 0 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Snapdragon® X Elite 9.771 ms 142 - 142 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile 6.505 ms 9 - 17 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) 9.062 ms 9 - 10 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 10.363 ms 8 - 12 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite For Galaxy Mobile 5.384 ms 9 - 17 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile 4.878 ms 10 - 20 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite 5.488 ms 142 - 142 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X Elite 14.96 ms 146 - 146 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Gen 3 Mobile 12.414 ms 4 - 15 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS8550 (Proxy) 16.212 ms 4 - 6 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS9075 18.591 ms 4 - 7 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite For Galaxy Mobile 8.621 ms 4 - 11 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite Gen 5 Mobile 8.046 ms 4 - 14 MB NPU
ZipformerEncoder PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X2 Elite 9.253 ms 147 - 147 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Snapdragon® X Elite 9.729 ms 3 - 3 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile 6.6 ms 2 - 10 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® QCS8275 (Proxy) 23.718 ms 3 - 11 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) 9.066 ms 3 - 5 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® SA8775P 10.992 ms 3 - 11 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® QCS9075 10.214 ms 3 - 7 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® QCS8450 (Proxy) 11.845 ms 2 - 12 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® SA7255P 23.718 ms 3 - 11 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Qualcomm® SA8295P 13.295 ms 3 - 8 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite For Galaxy Mobile 5.433 ms 3 - 15 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile 4.738 ms 3 - 12 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Snapdragon® X Elite 14.608 ms 1 - 1 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Gen 3 Mobile 11.568 ms 1 - 8 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8275 (Proxy) 27.668 ms 1 - 8 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8550 (Proxy) 15.779 ms 1 - 5 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Qualcomm® SA8775P 17.364 ms 1 - 10 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS9075 17.302 ms 1 - 3 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Qualcomm® SA7255P 27.668 ms 1 - 8 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite For Galaxy Mobile 8.509 ms 1 - 9 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite Gen 5 Mobile 7.924 ms 1 - 11 MB NPU
ZipformerEncoder QNN_CONTEXT_BINARY w16a16 Snapdragon® X2 Elite 9.366 ms 1 - 1 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Snapdragon® X Elite 0.262 ms 6 - 6 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Snapdragon® 8 Gen 3 Mobile 0.149 ms 0 - 6 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Qualcomm® QCS8550 (Proxy) 0.19 ms 0 - 8 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Qualcomm® QCS9075 0.299 ms 0 - 3 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite For Galaxy Mobile 0.134 ms 0 - 8 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Snapdragon® 8 Elite Gen 5 Mobile 0.132 ms 0 - 10 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX float Snapdragon® X2 Elite 0.079 ms 6 - 6 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X Elite 0.25 ms 6 - 6 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Gen 3 Mobile 0.157 ms 0 - 7 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS8550 (Proxy) 0.187 ms 0 - 8 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Qualcomm® QCS9075 0.253 ms 0 - 3 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite For Galaxy Mobile 0.132 ms 0 - 7 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Snapdragon® 8 Elite Gen 5 Mobile 0.135 ms 0 - 10 MB NPU
ZipformerJoiner PRECOMPILED_QNN_ONNX w16a16 Snapdragon® X2 Elite 0.083 ms 6 - 6 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Snapdragon® X Elite 0.246 ms 0 - 0 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Snapdragon® 8 Gen 3 Mobile 0.159 ms 0 - 7 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® QCS8275 (Proxy) 0.472 ms 0 - 7 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® QCS8550 (Proxy) 0.187 ms 0 - 1 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® SA8775P 0.324 ms 0 - 5 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® QCS9075 0.235 ms 0 - 2 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® QCS8450 (Proxy) 0.231 ms 0 - 9 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® SA7255P 0.472 ms 0 - 7 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Qualcomm® SA8295P 0.469 ms 0 - 5 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Snapdragon® 8 Elite For Galaxy Mobile 0.133 ms 0 - 13 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY float Snapdragon® 8 Elite Gen 5 Mobile 0.133 ms 0 - 10 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Snapdragon® X Elite 0.253 ms 0 - 0 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Gen 3 Mobile 0.151 ms 0 - 7 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8275 (Proxy) 0.438 ms 0 - 9 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS8550 (Proxy) 0.183 ms 0 - 1 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Qualcomm® SA8775P 0.883 ms 0 - 8 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Qualcomm® QCS9075 0.233 ms 2 - 4 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Qualcomm® SA7255P 0.438 ms 0 - 9 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite For Galaxy Mobile 0.135 ms 0 - 9 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Snapdragon® 8 Elite Gen 5 Mobile 0.135 ms 0 - 10 MB NPU
ZipformerJoiner QNN_CONTEXT_BINARY w16a16 Snapdragon® X2 Elite 0.176 ms 0 - 0 MB NPU

License

  • The license for the original implementation of Zipformer 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