All Products
Search
Document Center

Use customization tools for speech recognition

Last Updated: Oct 27, 2020

Intelligent Speech Interaction provides recognition customization tools to improve the effectiveness of speech recognition. If the speech recognition services cannot recognize unique words in your business field, you can use the hotword feature. If the basic models that are provided by Intelligent Speech Interaction cannot meet your business requirements, you can create custom models in the console to optimize the basic models.

For more information about how to train and manage custom models in the console, see Manage custom models.

Training corpus

The self-learning platform allows you to configure hotwords and create custom models.

  • Hotwords: You can configure hotwords to improve the effectiveness of recognizing specific words. A hotword can be the name of a person or a location or a business exclusive word. For more information, see Introduction.

  • Custom models: You can optimize basic models to improve the effectiveness of speech recognition in specific scenarios. If you want to recognize proper nouns and high-frequency words, you can obtain optimal results by using custom models. For more information, see Introduction.

Use cases

A redology research seminar is about to be held. To record the audio of the guest speakers, the host chooses Intelligent Speech Interaction for speech transcription. Developers register an Alibaba Cloud account and activate Intelligent Speech Interaction. To improve the effectiveness of speech recognition, the developers train a custom model on the self-learning platform.

  1. Select a basic model. Developers choose to use a universal model.

  2. Collect a training corpus. This seminar focuses on topics that are related to Dream of the Red Chamber. Developers crop the original text based on punctuation marks and store each sentence as a separate line in the training corpus.

  3. Train the custom model. On the self-learning platform, developers upload the training corpus and train the custom model based on the training corpus. This way, developers can use the trained model to recognize specific words in Dream of the Red Chamber, such as Jia Baoyu, to obtain optimal recognition results.