Supported WAV File formats

For those that wish to create their own WAV files, you will need to know about some specific constraints of exactly what the DAB does and does not support.

PCM. 8 or 16 bit linear (uncompressed) values, in either mono or stereo format. The maximum supported sample rate is 22.05kHz.

muLaw. This is a logarithmically compressed format which stores 8 bits per sample and expands to approximately 12 bits of precision. Mono and Stereo forms are supported. The maximum supported sample rate is 22.05kHz.

IMA/DVI ADPCM. Adaptive Delta Pulse Code Modulation essentially stores the difference from one sample to the next. Several different ADPCM methods and subsets are part of the WAV file specification, the one that the DAB supports is specifically this type, with 4 bits per sample, in mono format only. The maximum supported sample rate is 16kHz.

For the types that support stereo file playback, no attempt is made to “mix” the two channels together and the data stream for only the left channel will be audible.

If a WAV file is stored using an unsupported compression method, or at too high of a sample rate, the DAB will simply treat the file as a missing file and refuse to play it.

Things to consider when deciding file storage parameters

If the user wishes to provide for the highest fidelity playback, then the files should be stored as 16 bit PCM files. Using this storage method consumes two bytes of disk space per sample.

If the user wishes to conserve as much disk space as possible, then the ADPCM compression method is recommended. Using this storage method, two samples are stored in one byte of disk space.

The default speech libraries included with the DAB utilize muLaw compression for the TI speech synthesizer recordings, and ADPCM for the live recordings. Both types are sampled at 11.025kHz.

It was found that applying ADPCM compression to the TI synthesizer recordings resulted in objectionable artifacts when the files were played, but these same artifacts were not nearly as noticeable for the live recordings.

The practical limit of a narrowband FM voice channel is about 48dB Signal to Noise ratio, so what may sound objectionable in your recording studio, may be inaudible to the end user.

Sample rates higher than 11.025kHz will be of little practical use when the high frequency rolloff of a normal two- way radio is considered. In fact, the end user may find that recordings at 8kHz sample rates are quite satisfactory.

When making your own live recordings using “built-in” audio hardware in your budget desktop computer, you may find that recording at these lower sample rates results in muddy sounding audio, and come to the conclusion that even a 22.05kHz sample rate just isn’t enough. In this case, the real culprit is the poor performance of the low cost audio card in your computer. Here you have two potential solutions. Either record at a higher sample rate, such as 44.1kHz and then downsample to a lower rate, or buy yourself a professional grade audio card that doesn’t have these same performance limitations.

7KDAB, Version 2.0, 19-Aug-2007