Stereo Wave Support

Translator encounters many situations where wavesamples, in stereo, mono, or dual-mono formats, have the opportunity to be organized in a different way. Translator attempts to retain the original format whenever possible, except if it is more advantageous to be organized in a different way in the destination format. (The other exception is if it has not been coded yet. Dual-mono is usually easier to initially implement.)

Dual-mono, panned hard-left and hard-right, is technically the same as stereo (see below for the exceptions). Most hardware samplers use dual-mono configurations, with the exception of Akai S-5000/Z-Series, Emu and Kurzweil. (The Akai is actually the only one that use interleaves, and it uses .wav files anyway.)

Most software sampler formats deal allow stereo interleaved files, with the exception of MESA and Pulsar (which are Akai derivatives), and SoundFont.

Two hardware samplers that use DOS-type files, the Korg Triton and the Roland XV-5080/Fantom-S, do not use stereo .wav files, so when either of these are the destination, stereo sources are split, and dual-mono samples are left alone.

Translator provides two options (see Options - General Tab for more info) regarding how Stereo files are handled: Combine to Stereo and Combine Unconditionally. Combine to Stereo, when checked, will take a pair of mono files and combine them into one Stereo sample if the destination format allows it.

Combine Unconditionally refers to the check Translator makes on a sample pair if Combine to Stereo is checked. The check mainly focuses on loop points, but can refer to sample rate, tuning, or other factors that would affect the samples and their difference in playback. If Combine Unconditionally is checked, Translator will bypass this check and only qualify a pair of mono files as stereo on the most basic level, listed below:

• Akai: If two samples contain -L or -R as the last two characters of the sample name. Note: Akai usually puts -L or -R in the 11th and 12th character. (Exception: the S3000 "plain" uses [space]L and [space]R, interestingly enough.) Additionally, Pulsar screws this up by writing -L and -R as just the last two characters, regardless of name length.
• Roland: If two samples contain -L or -R as the last two characters of the sample name. (Actually, Roland uses the ASCii number 127 as the dash indicating stereo, and the regular ASCii number 45 as a dash within the name.)
• Ensoniq: If two samples use consecutive sample numbers, are on consecutive Layers and the Stereo Layer Link is ON on both layers.
• Kurzweil: 1) If two samples contain -L or -R as the last two characters of the sample name, or if the Sample is a legitimate Stereo Sample with two Sample headers.
• SoundFont: If two samples contain -L or -R, [space]L and [space]R, or (L) or (R), as the last characters of the sample name, or if the Link parameter of a sample is set. (This last function has not been accurately or actively supported by various SF programs or developers.)

Non-looped dual-mono waves, deemed as a possible stereo pair, are always combined even if the sample sizes are different, regardless of the Combine Unconditionally setting, with the trailing data of the shorter sample set to silence.

Again, dual-mono waves (panned hard left and right) are technically the same as stereo waves, with three notable exceptions:

• With most (if not all) platforms, streaming (that is, the procedure of playing back a wave) is less stressful on the playback machinery, thus reducing latency. (Nemesys has told us that it doesn't make a ton of difference on their units, but we will take the conventional wisdom and assume that stereo is better.)
• When editing (on most platforms), editing parameters on one side of a stereo wave edits the other side automatically. This is usually a good thing, unless you are getting creative and want to make changes to one side and not the other (The famous Distorted Reality CD-ROM's are great examples of this.)
• Stereo waves are easier to organize and streamline within an interface, since it reduces the number of waveforms within a structure.