20240038233. TRAINING A DEVICE SPECIFIC ACOUSTIC MODEL simplified abstract (SoundHound AI IP, LLC)

From WikiPatents
Jump to navigation Jump to search

TRAINING A DEVICE SPECIFIC ACOUSTIC MODEL

Organization Name

SoundHound AI IP, LLC

Inventor(s)

Keyvan Mohajer of Los Gatos CA (US)

Mehul Patel of Santa Clara CA (US)

TRAINING A DEVICE SPECIFIC ACOUSTIC MODEL - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240038233 titled 'TRAINING A DEVICE SPECIFIC ACOUSTIC MODEL

Simplified Explanation

The abstract of the patent application describes a method for configuring custom acoustic models for speech recognition systems. Developers can provide audio files with custom recordings to train a baseline model. The custom acoustic model can include custom noise to simulate real-world conditions. This model can be used as an alternative to a standard acoustic model based on device conditions or type. The speech recognition system can select the appropriate acoustic model for speech recognition tasks. The results can be provided to developers, including an error rate.

  • Developers can configure custom acoustic models by providing audio files with custom recordings.
  • The custom acoustic model is trained by tuning a baseline model using the provided audio files.
  • Custom noise can be added to the audio files to simulate real-world conditions during training.
  • The custom acoustic model is an alternative to a standard acoustic model.
  • The speech recognition system can select the appropriate acoustic model based on device conditions or type.
  • Speech recognition is performed using one or more acoustic models.
  • The results and error rate can be provided to developers through the user interface.

Potential Applications:

  • Speech recognition systems in various industries such as healthcare, customer service, and automotive.
  • Voice-controlled devices and virtual assistants.
  • Transcription services and voice-to-text applications.

Problems Solved:

  • Improves the accuracy and performance of speech recognition systems by allowing customization.
  • Addresses the challenge of adapting speech recognition to different device conditions or types.
  • Enables developers to train acoustic models with custom recordings and noise for more realistic speech recognition.

Benefits:

  • Increased accuracy and reliability of speech recognition systems.
  • Customization options for developers to improve performance in specific scenarios.
  • Better adaptation to different device conditions or types.
  • Potential for improved user experience and productivity in speech-based applications.


Original Abstract Submitted

custom acoustic models can be configured by developers by providing audio files with custom recordings. the custom acoustic model is trained by tuning a baseline model using the audio files. audio files may contain custom noise to apply to clean speech for training. the custom acoustic model is provided as an alternative to a standard acoustic model. a speech recognition system can select an acoustic model for use upon receiving metadata about the device conditions or type. speech recognition is performed on speech audio using one or more acoustic models. the result can be provided to developers through the user interface, and an error rate can be computed and also provided.