AI-NEWS · 2024年 8月 10日

ByteDance’s Doubao Model Supports Real-time Voice Calls: Can Be Interrupted Anytime with Instant Responses

ByteDance's Doubao Model Supports Real-time Voice Calls

  1. Announcement and Features:

    • ByteDance has introduced a new feature for Doubao's large model supporting real-time voice calls.
    • This feature allows voice calls to be interrupted anytime with instant responses.
  2. Technology Integration:

    • The solution combines Volcano Engine's platform with Doubao’s voice recognition and synthesis models.
    • Simplifies the conversion process between voice and text, ensuring efficient data collection, processing, and transmission.
  3. Key Technologies:

    • Uses Volcano Engine RTC and audio 3A processing technology to eliminate the "double talk" phenomenon.
    • WebRTC transmission network enables ultra-low latency and stable, reliable global real-time audio and video services.
  4. Flexibility and Accessibility:

    • Provides diversified access solutions like self-integration and WebRTC standard protocol-based transmission networks.
    • Tailored to meet the specific needs of various enterprises.
  5. Industry Applications:

    • Already deployed in leading domestic AI virtual character chat applications, enhancing interactive experiences.
    • Continues to offer high-quality audio and video capabilities and AI innovations to help enterprises.
  6. Future Directions:

    • Ongoing commitment to providing cutting-edge AI real-time audio and video solutions.
    • Aims to support enterprise innovation in AI-driven interactions.

Source:https://www.aibase.com/news/10955