Current speech and audio processing algorithms do not explicitly process sounds the way they are perceived by the human ear. Loudness estimation algorithms process sounds consistent with human auditory perception.

A fast technique for loudness was developed without sacrificing performance. The algorithm provides access to different intermediate quantities (excitation pattern, specific loudness, Instantaneous loudness, short-term and long-term loudness). Depending on the needs of an application, different quantities can be estimated in a computationally efficient manner.

Brief Overview:

Audio processing applications such as rate determination, bandwidth extension, compression, and noise reduction make use of loudness metrics. Most loudness models are computationally expensive and often not suitable for real time applications. Here, we present a low-complexity loudness estimation algorithm for both steady and time-varying sounds. The model computes an estimate of the excitation pattern by simultaneously pruning the frequency components and detector locations. For time-varying sounds, intensity patterns across successive audio frames are further exploited to reduce complexity by partial evaluation of the excitation pattern in select critical bands. Comparative results indicate that the proposed algorithm performs consistently well for different types of audio signals at a reduced complexity.

Complexity comparison

The complexity of the reference algorithm is shown in blue. The complexity with detector pruning is shown in purple. The complexity with detector and frequency pruning is shown in yellow.

Performance comparison

Additional Information

For additional information on our technology and its use in applications and their benefits, refer to the following documents.

- Harish Krishnamoorthi, Visar Berisha, Andreas Spanias, " A Low-Complexity Loudness Estimation Algorithm", Proc. of IEEE ICASSP-08, April 2008, Las Vegas, NV..

- Visar Berisha and Andreas Spanias, " Wide-band Speech Recovery using Perceptual Criteria", EURASIP Journal on Audio, Speech, and Music Processing, Vol 2007, Article ID 16816, 18 pages, August 2007..