|
[What exactly…] See below. [Space…] The first thing I checked, of course. My laptop: 398G. The machine at school: 980G. We use an external 1T hard drive and never fill it more than 200G just to avoid errors like this. But these recordings often
stop after 2 seconds, or 40 seconds. Hardly a storage issue. [How exactly…] We use Logic to record the CAF files. It’s an option in preferences.
We restart logic with a macro that very quickly presses space (stop), return (return to beginning), R (start recording). Logic starts a new batch of stereo pair CAF files and we split, then move the older
ones. As far as I understand, we could record endless CAF files, but that would be pretty inefficient once they get really large.
[How exactly are you splitting… ] I have predefined blocks. Every hour up to 7:30, then 7:30 to 10:00 (concerts). When the script runs it checks the time. Finds the previous block (e.g., if splitting at 8:02, it splits 7 to 8). We use SoX
in combination with the start time of the file to precisely cut each hour block (trim). They can then be recombined without losing any material. It’s extremely accurate. They are saved to another hard drive as aif.
[Is that a limitation… ] I appreciate the cynicism, but yes, that was the first thing we checked, letting Logic record till it stops, around 13 hours. There are a number of articles if you want to search, but we did try it. I think
the limitation is something stupid, like the number of measures it can accommodate.
[Limitation to CAF…] From Wikipedia, hundreds of years: “.caf container can contain many different audio formats, metadata tracks, and much more data. Like the RF64,
it is not limited to a 4 GB file size and a single .caf file can theoretically save hundreds of years of recorded audio due to its use of 64-bit file offsets.” That’s why we use it rather than AIF or WAV. We would have to split those up every 3 hours and that would create more problematic seams.
[How do you record 24/7…] I said 24/7, but we don’t go overnight because there’s no point, but we could. We let Logic run about 10 hours, then reset it using the space, return, R macro. We lose 400ms with that operation, but we have
two redundant machines and if there is a concert or a take at that split we go to the other machine (which splits at a different time). But we’ve also found we can edit the 400ms gap.
[You would be better off… ] EXACTLY! That’s what I’m trying to do. I’m partway there, but I’m not going to rely on it if it stops after 20 seconds.
[0 * * * * $... ] I tried this and the trim didn’t have any affect I could see. It simply recorded and continued recording. (I used trim 10 rather than 3600.) If I could get sox to record 4 stereo files split at the hour and
not stop after 20 seconds I’d switch tomorrow. Something like three terminal windows, each with a command: % rec -c 2 1 hall1.aif trim 3600;
% rec -c 2 3 hall1.aif trim 3600;
% rec -c 2 5 hall1.aif trim 3600;
Where the second variable is the starting channel (2 channels starting at 1, then 3, then 5). And the 3600 splits the file at the sample level but also ***starts a new file without killing the old one. I just
did a test and was able to do two windows with a recording. That would be ideal. I just don’t know SoX that well.
[So why everything all the time… ] Because we don’t know when there will be someone in a room practicing, doing a session (they’ve come to rely on this system), or recording video with a phone then they pull our higher quality audio to sync
it. We’ve been doing this for 15 years and it’s revolutionized how students and faculty work. They can get a recording of anything they do anytime. Some students just see an empty room or hall, walk in, play for a half hour and request the recordings. Organists
practice at 6am. If they nail a Bach, they have a recording. (Or maybe they just want to listen to how it sounded with the acoustic curtains up or down.) FWIW, we’ve won a few awards. And we NEVER miss a concert. Even if it was changed to a different room
or time at the last minute. [Long pause… ] I actually worked with Roni Music and talked them into allowing 10 minutes of silence before shutting off. We also put the threshold at -64 dB. After a few years it was pretty well fine tuned. But as I said,
OS X’s crack down made use rethink the whole system (and I spent a few weeks trying to get SoX to work), then moved to Logic, and it works pretty well. It’s been flakey lately, so I’m looking into SoX again. About once a year I look for another method and
we’ve even pitched this system for commercialization with the IP people at the school (everyone loves it, they are just used to millions, not hundreds of thousands, which they say it would make).
[Is some timestamp… ] No, we just use the creation date and calculate from there. It’s very precise.
[Human operator… ] No. But human operators interact with the system daily, checking for quality or errors, or if Logic crashes (about once every 4 months, it sends an email to three engineers if something looks wrong, like a
change date longer than 2 seconds on the audio files folder) or we get digital faults from the interfaces running too long (about twice a year, if we forget to restart them about once a month), or sometimes for several hours. For example an important concert
we sit and watch monitor it or if we are hired to do a session and they want someone there monitoring the entire time, then they just sit down and monitor the recordings that are already happening. (But also, they may make modifications, like put two mics
on the floor and change the inputs for the particular session.) I even have a few faculty who understand the system well enough to go into a hall (a wind quartet this last week), do a screen share to check the recordings and levels, start takes, work for 6
hours, pull the files, go home. Engineer cost: $0. Other schools: $600. Times four days. Us: $0. Other schools: $2400. Just to have someone sit there and watch the flashing lights.
[The biggest advantage…
starting another] Yes, exactly, but I need it to do 8 stereo pairs to separate files on two Mac Minis but most important, split the files at the sample level, no gaps. I’m not sure command lines can be that precise. Unless
there is a split after x minutes. If so, that would solve it. I tried the rec file.aif trim 10, but didn’t work the way I would expect (splitting every 10 seconds, creating a new file).
[rec -V5…] Looks like buffer overrun. This stopped after 25 seconds.
minion-2-recording:~ minion$ ~/sox/rec -c 8 -b 16 -r 48k /Volumes/CAF3/test.caf /Users/minion/sox/rec WARN formats: can't set 8 channels; using 18 Input File
: 'default' (coreaudio) Channels
: 18 Sample Rate
: 48000 Precision
: 32-bit Sample Encoding: 32-bit Signed Integer PCM In:0.00% 00:00:25.69 [00:00:00.00] Out:1.23M [
|
]
Clip:0
/Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. /Users/minion/sox/rec WARN coreaudio: coreaudio: unhandled buffer overrun.
Data discarded. In:0.00% 00:00:25.71 [00:00:00.00] Out:1.23M [
|
]
Clip:0 Done. |
_______________________________________________ Sox-users mailing list Sox-users@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/sox-users
