Voice Interaction System Latency Reduction via STT-TTS Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI systems face challenges in enabling low-latency voice communication and integrating with legacy systems, which results in high latency, increased network traffic, and difficulties in handling legacy data formats.
Innovation Solution
The system employs a bidirectional communication connection to receive audio data from users, convert it to text using STT services, generate AI prompts, and send them to AI models for response. The system then converts the response into speech using TTS services and sends it back to the user at a target bitrate, minimizing latency and optimizing network traffic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If voice communication is implemented with AI models using traditional systems, then the system can process voice data, but the latency is high and network traffic is increased
Solution Approach 1:
The patent segments the voice communication process into distinct processing stages: audio reception, STT conversion, prompt generation, AI processing, TTS conversion, and speech delivery. Each stage processes only the necessary data for its specific function, reducing overall latency and network traffic by avoiding redundant data transmission across the entire system.
Solution Approach 2:
The patent introduces intermediary services (STT and TTS) that act as mediators between the voice input/output and the AI model. These intermediaries convert voice to text and text to speech, allowing the AI model to process only text data rather than raw audio, significantly reducing the computational burden and latency while minimizing network traffic.
2Adaptability or versatility
If the system integrates with legacy systems, then it can accommodate existing infrastructure, but data formatting difficulties and increased complexity arise
Solution Approach 1:
The patent changes the data format parameter from raw audio to text representation through STT conversion. This parameter change allows the system to interface with legacy systems that handle text data more efficiently, accommodating existing infrastructure while reducing the complexity of processing and formatting requirements.
3Reliability
If complete voice processing is performed before response delivery, then accurate AI responses are generated, but the latency increases significantly
Solution Approach 1:
The patent performs preliminary actions by converting voice to text using STT before the AI model processes the data. This preliminary conversion allows parallel processing where the AI model can begin generating responses while the system prepares TTS conversion, reducing overall processing time while maintaining response accuracy through complete voice analysis.
Data Source
AI summary
This disclosure describes systems, computer readable media, and methods for establishing a user call on a telecommunications network, the call including a user phone number, and establishing a bidirectional communication connection with the user. Audio data is received from the user and sent to a STT service. Text data is received from the STT service, the text data representing the audio data. The AI prompt can be sent to an AI model, from which a text response is received. The text response can be parsed into one or more response statements, which are sent to a TTS service. A stream of speech data can be received from the TTS services and converted to a format suitable for the telecommunications network. A target bitrate for the converted speech can be determined, and the converted speech can be sent to the user at the target bitrate.