From now on SoundExpert is ready to accept wider class of devices/technologies for testing and rating. Hardware audio equipment for example. This became possible owing to development of precise time warping procedure which eliminates various time distortions of audio signals usually caused by AD/DA conversion, filtering, inaccuracy of clock rate, etc. Being inaudible and practically immeasurable such time/phase inconsistency affects greatly the difference signal used for artifacts amplification in SE testing scheme. Fortunately all software codecs produce time accurate output while most of other devices/processings usually have constant or variable (wow-and-flutter) time shifts. Now we can precisely adjust time grid of an audio signal to time grid of reference signal. Thus difference signal contains only shape differences regardless of time squeezing/shrinking.

The first device that will be added to SE rating system using this kind of “time treatment” is hardware codec ATRAC SP (Type-R). It has no software implementations and used in MD players/recorders/decks. Recording/playback of ATRAC SP streams as many other real-time processes is affected by inaccuracy of clock rates. Resulting digitally recorded streams usually have small variations of their time scales, not audible but sufficient to spoil difference signal completely. So ATRAC SP is a good candidate for testing our new time warping technology.

We thank Lee Keywood who not only suggested inclusion of this codec for benchmarking purposes but also encouraged us to develop the time warping procedure. He carefully made on his MD equipment tens of recordings that we used for testing and tuning of our code.

This successful collaboration convinced us in possibility of preparing testing material outside SoundExpert by volunteers. Corresponding methodology of selection and verification of submitted output samples was also developed and now needs to be tested in practice.

For the purpose we ask volunteers who have MD equipment with ATRAC SP (Type-R) capability to record digitally SE test file and then record it back to PCM format digitally as well. This is in short. In detail:

1. Download a single test file - [link deleted as outdated] (WAV, 44100, 16, stereo, duration 5:15, size 29.7 MB). It consists of three identical groups of nine SE test samples (with trailing samples). Three groups are necessary to verify the consistency of recoding. The test file can be played back on computer directly or by mounting virtual Audio CD. Also it can be burnt as real Audio CD and played back on CD/DVD player. CUE-sheet for burning/mounting applications is included.
2. Using S/PDIF, TosLink, AES, USB or other digital interface record the test file to MD (Type-R) as a single track.
3. Playback the recorded MD track through digital interface and record it back to computer or other recording equipment capable of recording PCM 44100. If you record in 32bit format, leave it as is to avoid additional processing.
4. Archive or losslessly compress resulting WAV/AIFF 44100 PCM file and upload it to some file hosting service.

As we never did this before we don't know how many such recordings will be sufficient (this is another goal of the experiment). We'll process all received files with our time warping software to eliminate wow-and-flutter and other time distortions. For every file submitted we'll upload corrected (warped) versions along with difference signals. So the whole experiment will be openly covered on SE site and could be discussed on Sony Insider forum.

The most consistent and representative recording will then be added to SE rating system as ATRAC SP (Type-R) device under test. Although this format is not developed anymore it is still used for home and semi-professional recording. It will serve as a benchmark for SE rating scale, let alone the possibility to compare it to other audio compression algorithms at various bit rates. Also if this experiment is successful the above scheme could be used for adding other audio equipment to SE rating system.

Thank you in advance,
SE Team.