This topic lists the frequently asked questions (FAQ) about the self-learning platform.
The FAQs for speech recognition are divided into the following categories:
Features
Effect Class
Features
What are the call limits for self-learning models?
For more information, see Call limits.
What is the difference between configuring a self-learning model in the console and using POP API?
For more information, see Differences between configuring a self-learning model in the console and using POP API.
How do I add hotwords in the console?
For more information, see Create hotwords.
How do I set the weight of generic hotwords in the console?
The console does not support adjusting the weight of generic hotwords. You can adjust the weight using the POP API. For more information, see Create a self-learning model using POP API.
How do I create a hotword vocabulary using POP API?
You can use the POP API on the client to train business-specific hotwords without using the console. You can perform operations such as creating a vocabulary (CreateAsrVocab), retrieving a vocabulary (GetAsrVocab), updating a vocabulary (UpdateAsrVocab), deleting a vocabulary (DeleteAsrVocab), and listing all vocabularies (ListAsrVocab). For more information, see Create a self-learning model using POP API.
Is there a limit on the number of hotwords?
A single vocabulary can contain a maximum of 500 hotwords. Each account can create up to 10 vocabularies. If the number of your hotwords exceeds the limit, you can use the self-learning platform to train a custom language model.
Are there other ways to optimize recognition performance besides using hotwords?
You can use the language model customization feature of the self-learning platform to perform text optimization.
Performance
How do I optimize training corpora for a self-learning model?
For more information, see Optimization suggestions.