|
-
Based on computational neurophysiologic models of what the brain actually interprets
(Non-psychoacoustic model) - Provides a next generation bit-allocation model based on a unique understanding of human neurophysiology
- Only music-specific codec specifically engineered for mobile networks to deliver high quality downloads and mobile streaming/radio applications
- Competitive alternative to AMR-WB+ and eAAC+ product lines
- Lowest bandwidth, high quality codec for mobile music/radio streaming and downloading
- Optimized for ARM based processor
- Lowest complexity, RAM and ROM requirements
-
The Neural Encoding Model (NEM™) including encoding functionality available on a licensing basis
Neural Encoding Model (NEM)
BrainMedia has pioneered the translation of the most recent and leading edge neurophysiology into computational models that can be programmed into codecs (a field called “computational neurophysiology”). Because of this pioneering position, BrainMedia was able to build a blocking patent portfolio related to these breakthroughs. The result is the new BrainMedia Neural Encoding Model (NEM™). This new codec identifies the 1-3% of audio signals that is actually represented in the brain (cortex). After music passes through the inner ear (where traditional codecs work), it is further filtered before it reaches the human cortex. It is this highly processed and filtered signal that the brain actually hears
After music passes through the inner ear (where traditional codecs work), it is further filtered before it reaches the human cortex. It is this highly processed and filtered signal that the brain actually hears.
Music Encoding Breakthrough
New performance gains in audio codec quality will only come from breakthroughs understandings of how the brain does and does not process sound. All standardized and de facto standard codecs are based on a primitive understanding of human sound perception (called “psychoacoustics”) pre-dating the recent revolution in auditory neurosciences. With advent of new and powerful experimental tools, academic researchers at leading institutes are unraveling for the first time the details of what the human brain can actually hear, and why.
Not Neural Network-Based
NEM is based on the actual neurophysiology of the human brain and is distinctly different than, for example, the field of “neural networks” which does not benefit from any insight into how the brain actually processes sound. The core of BrainMedia’s audio encoder is our patented Neural Encoding Model that reveals which abstract nuances of any audio signal are processed by the brain and represented in the cortex. By comparing the neural representation of the highly compressed audio signal to the representation of the original audio, an advanced statistical analysis calculates the “perceptual distance” between the compressed and original audio.
How it Works
The NEM codec minimizes the perceptual distance of the compressed audio from the original audio; a small perceptual distance implies that as far as the cortical representation is concerned, the person is truly enjoying the perception of the original signal. This results in codec audio performance since the NEM model guides the precise placement of the tiny fraction of “bits” that are necessary to represent the audio with near zero perceptual distance from the bit-rich original signal.
NEM and NEMx
-
Lowest bit-rate, highest quality codec for mobile music applications including:
- Competitive alternative to AAC+ at 16 to 32 kpbs
- NEMx – NEM with new spectral extension support delivering full Stereo output at 14-24 Kbps with better quality than eAAC+ (AAC+ version 2.0) or AMR-WB+(Portions ©2007 ATC Labs. All Rights Reserved.)
-
Protected by US and international patents and not subject to any of the current royalty and IP issues associated with many psychoacoustic technologies
- Streaming radio
- Mobile game soundtracks
- On-device, rich media applications
The BrainMedia Advantage
The introduction of the eAAC+ and AMR-WB+ codecs demonstrated to the market that the performance of an audio codec can be augmented at low bitrates by using techniques based on “source models”. Source model techniques such as “spectral extension” and “parametric stereo” can be used to approximately recreate portions of the audio signal that were not actually encoded by the “baseband codec”.
For example, in eAAC+, the AAC codec is used to encode only a portion of the lower frequencies (baseband audio), and spectral extension(the “+” part) and parametric stereo (the “e” part) are used to reconstruct the omitted audio. A low bit rate is achieved by using the baseband as a source of “hints” on how to approximate the missing upper bands, as well providing additional compact hints, in lieu of sending the high bit rate AAC upper bands.
- BrainMedia’s NEMx codec is the superior answer to eAAC+ and AMR-WB+
- Proven to be greatly superior to AAC in independent audio quality comparisons (e.g. MOS Testing)
- Capable of encoding more frequencies with fewer bits at a higher quality than AAC and AAC+
- Spectral extension and parametric stereo adopted in NEMx is derived from the most recent advances in these fields since “+” was introduced, and provides more perceptually accurate reconstruction of the upper bands for fewer bits.
The result is a superior user experience based on substantial advances in quality and lower bitrates over either eAAC+ or AMR-WB+