ATC Labs : Audio Post Processing

Low bit rate audio coding often results in the loss of a number of key audio attributes such as audio bandwidth and stereo separation. Additionally, there is also typically a loss in the level of details and intelligibility and/or warmth in the signal. Due to the proliferation, e.g. on Internet, of low bit rate audio coded using a variety of coding schemes and bit rates over which the listener has no control, it is becoming increasingly attractive to incorporate processing tools in the player which can ensure a consistent listener experience. Various novel post processing tools have been developed by ATC Labs which feature low complexity and easy to deploy techniques.

Technology Highlights :

Stereo Enhancement

Stereo signals with limited stereo are often found in the realm of compressed audio (i.e. music in compressed low bit rate formats) or in a FM receiver during instance of poor coverage (e.g. when the Side channel is lost or is of poor quality). The technique developed by us has the following features:

Enhances either a monophonic signal or a stereo signal with limited stereo image.

For signals with limited stereo optimum use of available stereo cues is made in generating the expanded stereo image.

Fully adaptive and based on a detailed time-frequency analysis of the signals and applies a combination of level and phase cues to achieve stereo separation and high image stability

Blind Bandwidth Extension

Low bit rate coded material available on the Internet is typically coded at relatively low bit rates, e.g., 48 kbps whereby the codec preserves only about 8 kHz of audio bandwidth. Restoration of higher frequency components prior to playback can therefore significantly enhance the listening experience. The advantage of a good blind Bandwidth Extension scheme is that it can be applied to degraded material over which the player has no control. ATC labs has developed a novel approach to blind bandwidth extension that provides a consistently superior sound quality across a wide range of audio material and exhibits a lot of detail in the regenerated high frequency component.

Extension of our patent pending Fractal Self Similarity Model (FSSM) bandwidth extension techniques.

Bandwidth extensible for audio starting with 5 kHz audio bandwidth

Automatic Noise Removal

The Automatic Noise Removal platform from ATC Labs consists of several noise removal and signal enhancement modules which may be used either independently or in conjunction with each other. Algorithms are fully automatic and signal adaptive (in other word these do not require manual selection of parameters such as noise profile, etc.) These may be used to enhance noisy audio often found in conventional radio receivers and some of these modules also work well to enhance signal affected by low bit rate coding artifacts. The noise removal tool consists of two key algorithms:

A novel wide-band Adaptive Noise Removal Module with Primary Sound Preservation (NRM)

NRM removes wide-band background noise with the help of adaptive filtering techniques.

Preserves primary signal sound by performing a detailed harmonic analysis of the signal and by utilizing our core expertise in perceptual modeling and accurate signal analysis and synthesis, removes the primary signal components from the signal prior to the step of noise-removal preventing damage to the main signal sound due to the cross affect of the noise removal algorithm (a key problem with most if not all current commercially available noise removal products)

NRM continuously and automatically updates noise profile/statistics with the help of novel voice activity detection (VAD) algorithm making the noise removal process fully automatic that does not require any human intervention.

A multi-band Temporal Envelope Enhancement Processor (TEEP)

A signal enhancement approach to reduce the impact of noise. In particular it attempts to enhance the temporal envelope coherence between the high frequency and low frequency signal components making the audio more stable and focused.

Also offers modules for other form of audio enhancements such as details enhancement and attack enhancement

Publications :

Raghuram Annadana, Harinarayanan E.V,Deepen Sinha, and Anibal Ferreira, “A Novel Audio Post-Processing Toolkit for the Enhancement of Audio Signals Coded at Low Bit Rates”, in the preprints of 123^rd Convention 2007 Oct 5-8, New York, USA...Pdf

Harinarayanan E.V, Deepen Sinha, Shamail Saeed and Anibal Ferreira “A Novel Automatic Noise Removal Technique for Audio and Speech Signals”, in the preprints of 123^rd Convention 2007 Oct 5-8, New York, USA...Pdf