Clan/mp3 timecoding update

[David Nathan]

Dear Rachel
We (below) also agree with the advice not to use MP3 as a recording format. The main problem is that MP3 involves lossy compression - ie you would be recording less accurate sound data than you could otherwise be doing. This is therefore simply not the best way to approach data collection. In turn, this has implications for preservation (archiving) and for purposing of the recordings for other useful products (e.g. involving the reconversion problems Linda identifies). 

Choices affecting recording quality are relative and format is but one of them. In the case of MP3, we'd also be worried that a fieldworker might be encouraged to use an inferior MP3 dictation recorder, or use MP3 merely as a way to get more sound onto a flash card (instead of preparing a proper transfer strategy). In addition, compressions vary - a high bit rate MP3 would be much preferable to a low one, and other compressions, eg WMA or Minidisc ATRAC, are better than MP3 (although MD carries its own disadvantages of transfer). Finally, some recording decisions, such as selection and usage of microphones, or recording location, can have a very much greater effect on your recordings than format choice. 

Re archiving, at ELAR, if sound data is only available 'natively' as MP3 we will archive those MP3 files. We would be particularly concerned about the case where fieldworkers record in MP3 and then convert the sound files to WAV - this not only bears all the disadvantages mentioned here and in Linda's message, but also (if not documented) hides the processing history of the data, and denies the archive the opportunity to make the 'best possible' conversion from MP3 (now or in the future). Therefore the 'archival format' formulation taken in isolation can be very misleading. While there are licence issues in relation to MP3 we would not at this stage class it as a proprietary format in the same sense as ATRAC or MS Word.
David Nathan, on behalf of ELAR and HRELP at SOAS