Digital sound processing Part 1
Every year, computer technologies are getting better and better, including software designed for professional audio data processing.
Microsoft took a big step forward with the development of the DirectX programming interface designed to simplify writing programs to work with graphics and sound.
Digital processing is subdivided into:
Linear processing occurs in real time and requires fast response from the processor.
Offline processing is not limited by time, so any processor can be used. But the processing process can take several hours.
In this article, we will consider aspects of digital audio processing such as compression, AudioMPEGLayer3 (MP3) technology, MP3 digital audio format, psychoacoustic modeling, adaptive encoding, digital audio storage, digital audio carriers.
1. Compress digital audio
With digital encoding, sound and video can be brought to the viewer, significantly reducing stream or bandwidth, and with upgrading computer technologies, known compression methods become cheaper and newer ones are becoming increasingly in demand.
Compression is carried out according to several rules:
If you can’t compress the data, try to do without it.
When compressing, use the lowest compression ratio.
Avoid compressing already compressed data.
Compression can be used to synchronize audio and video streams.
Use data without noise when compressing.
After compression, the possibility of data transmission errors increases.
2 AudioMPEGLayer3. MP3 digital audio format
MPEG (MotionPicturesExpertGroup) is the name of a working group established by the International Organization for Standardization and the International Committee on Electricity (ISO / IEC) to develop standards for video and audio compression. MPEG itself defines audio and video formats that use lossy compression, as well as the operations performed by MPEG decoders.
The MP3 standard is a very lossy audio compression scheme, the full name is MPEG-1 Layer3 (sometimes only MPEG Layer 3).
MP3 uses spectral clipping, based on the psychoacoustic model. The audio signal is divided into equal segments, each of which, after processing, is recorded in its own frame (frame). Spectral decomposition requires continuity of the input signal. Two or more peaks located next to each other are replaced by an averaged one. After spectral removal, mathematical compression and frame packing methods are applied. Each box can have multiple containers, allowing you to store information about multiple streams (per channel).
