DSP (Digital Signal Processing Technology) is an emerging discipline that uses computers or special processing equipment to process signals in digital form to obtain a signal form that meets people's needs. It involves many disciplines and is widely used in many fields. Video applications are the most extensive.
Yealink Audio Call Technology
There are many DSP implementation methods that are widely used in the market, and two are mainly listed here. One is to use a general-purpose programmable DSP, that is, a non-dedicated DSP chip. Compared with single-chip microcomputers, DSP chips have software and hardware resources more suitable for digital signal processing, and can be used for complex digital signal processing algorithms. The second is to use a dedicated DSP chip.
In some special occasions, the required signal processing speed is extremely high, and it is difficult to realize it with a general-purpose DSP chip. This kind of chip implements the corresponding signal processing algorithm in the chip with hardware without programming.
Most of the audio echo cancellation algorithms are based on DSP, and rarely on Windows computers. The main reason for this is real-time and high efficiency.
DSP chip resources are limited and the echo cancellation algorithm requires relatively large resources. Take a video conference system as an example. A large-scale conference room can generate echoes of more than 512ms. To eliminate such long-latency echoes, even if calculated at a sampling rate of 8k Hz, the length of the adaptive filter W(n) will reach 4096. In this way, on the one hand, a very large storage space is required to store W(n). On the other hand, the amount of calculation required for updating W(n) is also doubled, and the efficiency of traditional adaptive filters is difficult to guarantee.
Yealink Audio Call Technology
Currently DSP-based echo cancellation algorithms are relatively mature, and there are also a number of specialized algorithm/chip companies on the market that can provide optimized DSP-based software echo cancellation modules, such as Yealink network audio and video. The system products all use exclusive intelligent noise canceling chips.
Related article: "Comparison of Advantages of Hardware Video Conference and Software Video Conference"
The latest application areas of echo cancellation technology are various VoIP applications based on the Windows platform, such as software video conferencing, VoIP softphones, etc. When the echo cancellation algorithm is applied to the Windows platform, compared with the traditional DSP platform, it brings both advantages and new difficulties. High efficiency is no longer a problem on the Windows platform. Now PCs have abundant CPU resources and massive memory resources, and even complex echo cancellation algorithms can run freely. However, the new troubles outweigh the benefits.
First of all, the Windows platform is a non-real-time platform, and audio collection and playback are also non-real-time for the echo cancellation algorithm. Under the Windows platform, it is difficult for an application to directly control the acquisition and playback of the sound card at the bottom layer, and what is obtained is a non-real-time audio stream, which brings about the synchronization problem of the acquisition and playback of the audio stream.
In fact, in general VoIP software, receiving the other party's voice (remote voice) and transmitting it to the sound card for playback is performed in one thread, while collecting local voice (near-end voice) and transmitting to the other party is performed in another thread. . The Acoustic Echo Cancellation (AEC) algorithm requires the data in the playback thread as a reference while performing echo cancellation on the near-end voice. It is very important to synchronize the data in these two threads, because if they are slightly out of synchronization, the adaptive filter in the acoustic echo cancellation algorithm will diverge. It's hard to tell.
From the above analysis, due to the non-real-time nature of the Windows platform, the echo cancellation technology based on the Windows platform is much more difficult than the DSP platform. The echo cancellation technology based on the DSP: the chip resources are small, the real-time performance is high, and the near-end voice is easy to realize Synchronization of the signal and the far-end reference signal; Windows-based echo cancellation technology: resources are relatively abundant, and the audio processing algorithm can be easily run, but the real-time performance is insufficient, and it is difficult to synchronize the near-end signal and the far-end reference signal, which makes it difficult to achieve The function of echo cancellation.
Related reading:
" Key Technologies and Core Advantages of VCS Video Conference Business: Interpretation of Yealink 's 2018 Financial Report ", "Key Technologies and Core Advantages of SIP Phone Interpretation of Yealink's 2018 Annual Report"