Microsoft's Embedded Audio Codec (EAC) is an experimental audio coder offering fine-grain bitrate scalability. The coding algorithm is described in [1], which also compares the performance of EAC against an earlier version of the Scala codec.
The following single-channel sound files compare coding efficiencies of EAC and Scala codecs at two different bitrates. All files are approximately 6 seconds in length with 44.1 kHz sample rate; samples are presented both as compressed bitstreams and uncompressed (decoded) .wav files. For verification, the EAC bitstreams can be decoded using the EAC player, while Scala bitstreams can be decoded using the demonstration decoder software available from this site.
Signal | Original file (uncoded) |
Mono Bitrate | ||||
32 kbit/s | 64 kbit/s | |||||
EAC | SCALA | EAC | SCALA | |||
harpsichord | harp.wav | bitstream | harp32.eac | harp32.sab | harp64.eac | harp64.sab |
decoded wav | harp32_eac.wav | harp32_sac.wav | harp64_eac.wav | harp64_sac.wav | ||
female voice | voice.wav | bitstream | voice32.eac | voice32.sab | voice64.eac | voice64.sab |
decoded wav | voice32_eac.wav | voice32_sac.wav | voice64_eac.wav | voice64_sac.wav | ||
castanets | cast.wav | bitstream | cast32.eac | cast32.sab | cast64.eac | cast64.sab |
decoded wav | cast32_eac.wav | cast32_sac.wav | cast64_eac.wav | cast64_sac.wav |
While EAC performs well at lower bitrates for some signals including harpsichord, other signals such as female voice and castanets suffer from clearly audible coding artifacts at 32 kbit/s. At the higher rate of 64 kbit/s, our listening tests indicate lower average coding efficiency for EAC compared to the Scala codec. Reduced coding efficiency at higher bitrates is reflected in higher average transparency bitrates, measured for EAC at 110 kb/s (mono) and 168 kbit/s (stereo), compared to the Scala codec which achieves transparency at 80 kb/s (mono) and 132 kbit/s (stereo). Error-resilient coding is a further advantage of the Scala codec.
[1] J. Li, "Embedded Audio Coding (EAC) with Implicit Psychoacoustic Masking", ACM Multimedia 2002, pp. 592-601, Nice, France, Dec.1-6, 2002.