VoIP Speech Recognition Text-to-Speech Network Resilience
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communication systems face challenges in maintaining audio quality during VoIP calls due to network packet loss and jitter, leading to unintelligible audio, which previous solutions like FEC and jitter buffers cannot fully mitigate without introducing delays or trade-offs.
Innovation Solution
Implementing automatic speech recognition to convert user speech into text data, which is transmitted alongside audio, and using a user-specific text-to-speech module at the receiving end to synthesize audio when network conditions deteriorate, ensuring reliable communication by providing both real-time voice and text outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If forward error correction (FEC) is used to mitigate network loss, then audio reliability is improved, but significant packet loss bursts still result in unintelligible audio
Solution Approach 1:
The patent introduces text data as an intermediary representation of the audio content. When audio packets are lost or become unintelligible, the receiving device can reconstruct the audio using text-to-speech synthesis from the transmitted text data, which travels through a more reliable transmission channel
Solution Approach 2:
The patent creates a textual copy of the audio content and transmits this copy separately. This text copy serves as a backup representation that can be converted back to audio when the original audio transmission fails, ensuring audio intelligibility under poor network conditions
2Stability of the object's composition
If jitter buffer length is increased to tolerate jitter, then audio stability is improved, but transmission delay increases
Solution Approach 1:
The text data acts as an alternative pathway that bypasses the need for large jitter buffers. By having text representations transmitted alongside audio, the system can switch to text-to-speech reconstruction when audio timing becomes problematic, reducing the required jitter buffer size and associated delays
3Loss of information
If speech recognition and text-to-speech modules are added to handle poor network conditions, then audio intelligibility is improved, but device complexity increases
Solution Approach 1:
The patent divides the communication system into separate functional modules: speech recognition module at the transmitter, text transmission module, and text-to-speech synthesis module at the receiver. This segmentation allows each module to be optimized independently and enables selective activation based on network conditions
Solution Approach 2:
The system dynamically switches between using direct audio playback and text-to-speech reconstruction based on real-time assessment of audio quality and network conditions. This dynamic adaptation allows the system to maintain audio intelligibility while managing computational resources efficiently
Data Source
AI summary
A device for communicating with a remote device is disclosed, which includes a processor and a memory in communication with the processor. The memory includes executable instructions that, when executed, cause the processor to control the device to perform functions of establishing, via a communication network, a communication session with the remote device; capturing a speech spoken by a user and generating audio data representing the captured speech by the user; encoding the audio data for transmission to the remote device via the communication network; converting the audio data to text data representing the captured speech; and transmitting, during the communication session, the encoded audio data and the text data to the remote device via the communication network. The device thus can provide the text data representing the captured speech when a quality of the encoded audio signal received by the remote device is below a predetermined level.