QUALCOMM Incorporated (20240282327). SPEECH ENHANCEMENT USING PREDICTED NOISE simplified abstract
Contents
SPEECH ENHANCEMENT USING PREDICTED NOISE
Organization Name
Inventor(s)
Asif Mohammad of San Diego CA (US)
Fatemeh Alishahi of Los Angeles CA (US)
Maliha Arif Zain of Orlando FL (US)
Sajeer Fazil of San Diego CA (US)
SPEECH ENHANCEMENT USING PREDICTED NOISE - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240282327 titled 'SPEECH ENHANCEMENT USING PREDICTED NOISE
Simplified Explanation:
This patent application describes a device that can remove background noise from an audio signal, specifically focusing on speech.
- The device includes processors that analyze the input audio signal to identify speech and generate a predicted noise signal.
- This predicted noise signal is then subtracted from the input audio signal to produce a cleaner output audio signal.
Key Features and Innovation:
- Utilizes trained models to predict and remove background noise from audio signals.
- Specifically designed to enhance speech clarity by removing unwanted noise.
- Can be used in various applications where clear speech is essential, such as conference calls, voice recordings, and audio transcription.
Potential Applications:
- Conference calls
- Voice recordings
- Audio transcription
- Podcast recording
- Language translation services
Problems Solved:
- Eliminates background noise interference in audio recordings.
- Improves the quality and clarity of speech in various audio applications.
Benefits:
- Enhanced speech clarity
- Improved audio quality
- Better user experience in audio communication
Commercial Applications:
Potential commercial applications include:
- Audio recording devices
- Communication devices
- Speech recognition software
- Language translation services
Questions about the Technology:
1. How does the device differentiate between speech and background noise in the audio signal? 2. What is the accuracy rate of the predicted noise signal generated by the trained model?
2. How does the device handle complex audio environments with multiple sources of noise?
Original Abstract Submitted
a device includes one or more processors configured to obtain an input audio signal including at least first speech of a first person. the one or more processors are configured to generate a predicted noise signal based on processing of the input audio signal by a trained model. the one or more processors are configured to subtract the predicted noise signal from the input audio signal to generate an output audio signal.
