AI Abstract Data Transmission for Low-Bit-Rate Calls
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current codecs are insufficient for voice or video calls over networks with low bit rate limits, such as NR-NTN or IoT-NTN networks, leading to suboptimal performance.
Innovation Solution
Implementing artificial intelligence (AI) models to transform voice or video data into abstract data, which is smaller in size, and then synthesize the data at the receiving end using AI models to maintain call quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current codecs are used for voice or video calls, then the transmission quality can be maintained on conventional networks, but the performance becomes suboptimal on networks with low bit rate limits such as NR-NTN or IoT-NTN networks
Solution Approach 1:
The patent extracts only the essential features and semantics from voice or video data using AI models, transforming full-resolution media into abstract representations that contain only the most important information for reconstruction. This extraction approach allows efficient transmission on low bit rate networks while maintaining acceptable call quality.
Solution Approach 2:
The patent changes the parameter representation of voice/video data by converting it into abstract features and semantics rather than transmitting the original high-dimensional data. This parameter transformation enables the data to be efficiently compressed and transmitted over networks with low bit rate limits.
2Loss of information
If voice or video data is transmitted directly without transformation, then the data完整性 is maintained, but the data size is too large for low bit rate networks
Solution Approach 1:
The patent extracts only the essential features and semantics from voice or video data using AI models, transforming full-resolution media into abstract representations that contain only the most important information for reconstruction. This extraction approach allows efficient transmission on low bit rate networks while maintaining acceptable call quality.
Solution Approach 2:
The patent creates an abstract copy or representation of the original voice/video data that preserves the essential information while occupying significantly less space. The abstract data serves as a compressed representation that can be transmitted efficiently and then converted back to usable form at the receiving end.
3Productivity
If data is compressed to reduce transmission size, then the bit rate requirement is reduced, but the call quality may deteriorate
Solution Approach 1:
The patent extracts only the essential features and semantics from voice or video data using AI models, transforming full-resolution media into abstract representations that contain only the most important information for reconstruction. This extraction approach allows efficient transmission on low bit rate networks while maintaining acceptable call quality.
Solution Approach 2:
The patent changes the parameter representation of voice/video data by converting it into abstract features and semantics rather than transmitting the original high-dimensional data. This parameter transformation enables the data to be efficiently compressed and transmitted over networks with low bit rate limits.
Data Source
AI summary
A data transmission method is provided. The data transmission method may include the following steps. A transmitting apparatus may generate a voice data for a voice call or a video data for a video call. The transmitting apparatus may transform the voice data or the video data into an abstract data according to a first artificial intelligence (AI) model, wherein the abstract data includes information related to the voice data or the video data, and a size of the abstract data is smaller than a size of the voice data or the video data. The transmitting apparatus may transmit the transmitting apparatus, the abstract data to a receiving apparatus. The receiving apparatus may synthesize a synthesized voice data or a synthesized video data from the abstract data according to a second AI model. The receiving apparatus may play the synthesized voice data or the synthesized video data.