Speech Intent Classification and Slot Filling

Intent Classification and Slot Filling aims to not only recognize the user’s intention, but also detect entity slots and their corresponding lexical fillers. Below is an example:

Different from its counterpart in Natural Language Understanding (NLU) that takes text as input, here the model predicts the semantics directly from audio input.

The full documentation tree is as follows:

• Models
• Datasets
• Checkpoints
  
  • Loading Local Checkpoints
  • Inference
  • SLU Models
• NeMo Speech Intent Classification and Slot Filling Configuration Files
  
  • Dataset Configuration
  • Preprocessor Configuration
  • Augmentation Configurations
  • Model Architecture Configurations
  • Loss Configurations
  • Inference Configurations
• NeMo Speech Intent Classification and Slot Filling collection API
  
  • Model Classes
  • Mixins
• Resources and Documentation

Example of Speech Intent Classification and Slot Filling can be found here.

Information about how to load model checkpoints (either local files or pretrained ones from NGC), as well as a list of the checkpoints available on NGC are located on the Checkpoints page.

Documentation regarding the configuration files specific to SLU can be found in the Configuration Files page.