Audio Model with TensorFlow.js with Ping Yu

In addition, we will show that the deployed model can be fine-tuned and saved at client side, achieving recognition of words outside the original vocabulary. Thanks to the cross-platform nature of the browser, the deployment and fine-tuning can occur in supported web browsers on a wide range of devices.