Capital sound: the updated MP3 format will provide studio quality
MTUCI scientists have improved the MP3 file format. Now it allows you to maintain high-quality sound with a smaller file size and makes it possible to reduce the volume of music files by 20%. The method also allows you to maintain the quality of phonograms even when reducing the file size by half. The effectiveness of the method has already been confirmed by measurements and examinations, in which people noted that changes in sound are almost imperceptible even when recording at a reduced speed.
MP3 quality improvement Method
MTUCI scientists have developed a way to maintain digital audio quality without increasing file size. Moreover, you can use it to reduce the file size by 20%. Formats like MP3 are usually used to reduce the volume of audio archives, but they add new distortions. New algorithms solve this problem.: they retain quality even when the file size is halved.
This method allows you to maintain quality even when digital sound recording has disadvantages that are perceptible to the ear, unlike the analog method, where noises are less noticeable and a person hears them anyway due to the peculiarities of hearing.
— The MP3 format, despite its popularity, has disadvantages. For example, when digitizing an analog signal (pulse-code modulation, PCM) into digital at a speed of 250-270 Kbit/s, nuances are lost, the so—called attacks that determine the saturation of sound and its intelligibility, the university's press service said.
The new method improves MP3 quality even at 56 Kbit/s — the sound becomes as high-quality as with the basic 16-bit PCM (the most popular audio encoding). This is confirmed by measurements and surveys of people. Experts have also patented a "non-distorting companding" algorithm, in other words, a method for effective compression without distortion and noise during the conversion of an audio track from analog to digital.
— The method compresses the dynamic range to 10-15 dB, but the sound is transmitted without distortion. And even when using a 16-bit format, the quality is comparable to a 32-bit input signal, meaning the negative aspects of 16 bits are reduced," the MTUCI added.
Thus, this method preserves sound until the digital representation "breaks down" — unlike the analog one, where the noise is less and the person hears the sound anyway. Thanks to this, MP3 at 56 Kbit/s sounds as good as a standard signal. Studies confirm this: people hardly notice the difference, the scientists noted.
Who can use the new algorithm?
Digital sound is now used everywhere: on streaming platforms, in podcasts, audiobooks, video game voiceover, movies, videos, voice messages, video calls, smart speakers, voice assistants. It is used even in hearing aids, says Anton Averyanov, CEO of the ST IT Group of companies, TechNet NTI market expert.
— The problem is that quiet sounds in the digital are often lost. According to the old standards, up to half of the data just crashes, and to fix it, you need to increase the file size, which has a bad effect on users. This problem is very relevant, especially at low bitrates or in conditions of limited memory and bandwidth. Distortion manifests itself in aspects such as the loss of quiet sounds, noises, a "metallic" hue or blurriness. For example, in MP3 at 56-128 Kbit/s, up to 50% of data is lost, which significantly degrades the quality — the sound becomes flat, the details disappear," he said.
The proposed solution is optimal for recording studios, streaming platforms and compact devices with limited memory, the expert concluded.
For the successful application of this technology, it is necessary not only to introduce it on a large scale, for example, on streaming services, but also to improve the overall quality of sound-producing equipment. In general, modern mid-segment smartphones already have good sound, which means that the new format has prospects, said Svyatoslav Pegov, director of the Center for Sports Programming, Algorithmic Robotics, Cybersecurity and Esports at Ufa University of Science and Technology.
As IT specialist Ivan Kalmykov noted, the technology in the field of signal preprocessing can really be a breakthrough: for example, algorithms based on adaptive noise filtering can reduce the bitrate of streaming audio to 24 Kbit/s without loss of speech intelligibility.
— The prospect of replacing MP3 with more modern algorithms seems reasonable. Currently, such solutions as Google's SoundStream are already demonstrating 128 Kbit/s quality at 3 Kbit/s — these are good numbers. In my opinion, the research of scientists on the modeling of audio correctors lays the foundation for innovation, like Google's," the expert added.
Patented preprocessing and companding methods open the way to more efficient studio archives and high-quality audio streaming even at low bitrates, says Yaroslav Seliverstov, a leading AI expert at University 2035.
— The ability to transmit the quality of a 32-bit signal through a 16-bit stream is a tremendous achievement in conditions of limited resources. This is especially promising for mobile applications, cloud storage, and AI systems where volume and accuracy are critical. Such a development can be considered not only promising, but also system—forming for a new stage of digital sound," he added.
Of particular interest is the signal preprocessing technology, which can improve sound quality at low bitrates. This can be useful in online broadcasting, podcasting, and even in esports, where the clarity of the audio component is important without excessive traffic consumption. It is also promising to use new algorithms in studio work, for example, to compress multitrack soundtracks without losing subjective quality, said Daniil Arzhakov, an IT expert at the Department of Esports at the Faculty of Gaming Industry and Esports at Synergy University.
Such developments can become the basis not only for improving existing formats like MP3, but also for creating new audio encoding standards that take into account modern requirements for saving resources and a high level of perception. This is especially true in the context of the growing consumption of media content and the need to optimize the infrastructure for data storage and delivery, the expert summed up.
Переведено сервисом «Яндекс Переводчик»
