Tim Gillett wrote:So we used 24 bits and still many captures were digitally clipped...
Hugh Robjohns wrote:Adding my own voice to the concenaus just for the sake of clarity, the clipping described obviously had nothing to do with the chosen wordlength, and everything to do with utter incompetence and poor supervision/project management.
Of course the clipping had nothing to do with the wordlength. But I did go on to explain how I initially set up the gain staging. These were based on my own tests and measurements of the maximum saturated output of a Metal cassette. Then to that I added a record headroom of 5db. That is why while that system remained in place, which it did for a couple of months, we had no clipping whatsoever, across two separate capture installations and many cassette transfers.
Indeed the clipping had nothing to do with wordlength but neither did it have anything to do with my work or directives. Rather in spite of them. The clipping occurred for a different reason. Obviously I cant say much here about what then ensued and why, but basically what do we expect happens when fixed record levels which are already relatively high but quite safe, are increased significantly?
Hugh Robjohns wrote:I agree that 16 bit capture -- with competent input level control -- would have been adequate given the maximum dynamic range capability of the cassette format.
Perfectly adequate I would have thought? The one exception in the back of my mind was the possibility in the archive of a dbx encoded and decoded
cassette. It may still have been fine with 16 bit capture and careful record levels but we may never know for that project as AFAIK no such recording turned up in the archive of many cassettes.
Hugh Robjohns wrote:However, had it been my project today I would have captured in 24 bits simply because all modern interfaces do by default, and the concern about saving storage space really isn't a relevant issue for the production stage and hasn't been for decades.
Again this was a cassette digitisation project and a very large one at that. ISTR the ongoing storage requirements surprised the IT department.
Hugh Robjohns wrote:After post-production of the tracks I would have saved/stored the processed files in a 16-bit format for convenience and to reduce the long-term archive capacity required.
My understanding was that the 24 bit files would not just be used for the capture and then truncated down to 16 bit for posterity, but would be preserved as full 24 bit stereo wave files (24/48), along with the processed co masters.