In the traditional RTC (voice communication) scenario, the entire processing chain of the audio engine, such as 3A algorithm, codec, packet loss compensation and other technologies, only processes voice. In real-time interactive entertainment scenarios (such as host Lianmai, interactive podcasts, game voice, etc.), music, sound effects, etc. will be involved. When an audio engine with the goal of optimizing voice quality is faced with such scenarios, the sound quality after algorithm processing will be different. Great discount. For example, the frequency of double-talk problems in voice scenes is low. However, when the music is always playing, as long as the other end talks, double-talk will be formed. If the echo cancellation is not handled properly, it will cause great damage to the voice or music; For example, the noise reduction algorithm will inevitably damage the music when it encounters music. Therefore, in the real-time interactive entertainment scene, it is necessary to optimize the entire audio algorithm processing link of the audio engine to improve the sound quality in the real-time interactive entertainment scene as much as possible.

News Fast Delivery’s Expert Q&A (September 14th – September 20th) We have invitedXu XiaoyuThe teacher discusses with everyone aboutThe audio algorithm optimization direction of the entire link of the audio engine of the real-time interactive entertainment scene.

Issues that can be discussed include, but are not limited to:

• Audio capture
• echo cancellation
• Noise reduction
• AGC
• Packet Loss Compensation
• Codec
• Traditional algorithms combined with deep learning to optimize

or other aboutReal-time interactive entertainment scene audio engine The audio algorithm optimization direction of the entire link is relatedQuestions are welcome!

Guest introduction

Xu Xiaoyu, senior audio algorithm engineer in Lizhi, is proficient in digital signal processing, adaptive filtering, 3A algorithm, deep learning, etc. In Lizhi, he is responsible for creating high-quality RTC audio engines in interactive entertainment scenarios through traditional audio algorithms + deep learning.

github account:https://github.com/ewan-xu

In order to encourage active questioning, Lizhi Group will select from the questioners after the question and answer. 5 Lucky members gifted lychees2022annual tributeT.

The Q&A style of OSChina experts is consistent, and any discussions and trolls that are not related to the topic are not welcome.

The following welcomes everyone to comment on the optimization direction of the audio algorithm of the entire link of the audio engine of the real-time interactive entertainment scene.Xu XiaoyuIf the teacher asks a question, please reply directly to the question.