Audio processing
Monday evening was looking fairly blank, until Howard phoned me with a file restore problem. Windows ME is a pain. Not much else you can say really. Managed to get on the Waddo net, but not a lot to say really.
Tuesday evening's short visit to the studio to reboot a computer turned into a long session of playing with the kit. Tim suggested I should run a memory test on the studio machine, since it keeps falling over, so I thought I'd play some music to pass the time. Having plugged in my laptop to the mixer, I decided to use Cooledit to generate some tones, and I started to play with the compressor again. I've been trying to raise the "elbow" on the compressor for some time, because with the threshold too low, everything sounds compressed, and I'd rather let everything below -4dBU sound uncompressed. Because the threshold is so high, I need to set quite a high compression ratio to squish down the range between -4dBU and +12dBU to between -4dBU and +4dBU, which is not too high to cut slightly when feeding the audio into the stream encoder, which has a hard limit at 0dBA, and I generally like to have 0dBU line up with -6dBA, giving 2dB headroom. Discovered I have to do this with the Spectra effect turned off on the SCOM, as it shapes high freqencies, and measuring compression with a 1kHz tone becomes a bit of a nightmare with it turned on (and is possibly where I've gone wrong before)... no objection to turning it on after I've got the compression curve I want. I had a bit of a play with the time constants, but decided to leave them on auto as a reasonable compromise on the range of different kinds of music and speech programming. I've now got it so that with PPM5 peaks, we get PPM4 peaks on the return monitor, and PPM7 peaks result in PPM5 peaks on the return, with greater compression so that it still "feels" much louder, but the audio levels are pretty much under control. Not sure about setting it up to feed two streams though... we'll have to see about that.
All this work resulted in Tim installing shoutcast and an encoder on gateway, and we ran a few tests, and played with the config until I was happy that the capture level wasn't too high (actually, I had to set it very low, because the output levels are set high to allow reasonable throughput to the logging). The upshot is that I can now drive the mixer to PPM7, and while it sounds very loud, it's not too badly distorted, and regular driving to PPM5 sounds reasonably open, with minimum compression... which is how I like my audio. I guess if I had an AGC, I'd set it to drive the compressor a bit harder (maybe up 6dB). This has a knock on effect of improving some tracks that "play" the elbow rather badly. For example "I Know Him So Well", the low levels seem to be pushed down, and then you get a high peak, which goes too high after the compressor has responded to the low level. This causes the track's level to go all over the place. Running it 6dB up seems to resolve this issue, but resulting in an obviously compressed output. Can't win really.
Wednesday was a hard day at work, but at least I can work out for myself why something doesn't work these days. The evening trip to Grantham was uneventful, but received a SMS from Dave inviting me to have a look at some converted PMR kit. How that I've upgraded my ham license, I can now use this kit, so I took up this offer. Ended up drinking whisky until gone midnight...
...making this morning a bit of a slow start. Dragged myself swimming, and did at least 24 lengths, probably more, but I lost count. For some reason, they were piping Radio 1 into the pool area rather than Classic FM. That combined with someone swimming a splashy front crawl in the middle lane (maximum disruption to everyone else), made the morning swim more stressful than the sigmoidoscopy I had last year. Anyway, I now feel rather groggy, sort of mild migrain symptoms, so I'll try to catch a nap later to try to rectify things.
Onwards...
Tuesday evening's short visit to the studio to reboot a computer turned into a long session of playing with the kit. Tim suggested I should run a memory test on the studio machine, since it keeps falling over, so I thought I'd play some music to pass the time. Having plugged in my laptop to the mixer, I decided to use Cooledit to generate some tones, and I started to play with the compressor again. I've been trying to raise the "elbow" on the compressor for some time, because with the threshold too low, everything sounds compressed, and I'd rather let everything below -4dBU sound uncompressed. Because the threshold is so high, I need to set quite a high compression ratio to squish down the range between -4dBU and +12dBU to between -4dBU and +4dBU, which is not too high to cut slightly when feeding the audio into the stream encoder, which has a hard limit at 0dBA, and I generally like to have 0dBU line up with -6dBA, giving 2dB headroom. Discovered I have to do this with the Spectra effect turned off on the SCOM, as it shapes high freqencies, and measuring compression with a 1kHz tone becomes a bit of a nightmare with it turned on (and is possibly where I've gone wrong before)... no objection to turning it on after I've got the compression curve I want. I had a bit of a play with the time constants, but decided to leave them on auto as a reasonable compromise on the range of different kinds of music and speech programming. I've now got it so that with PPM5 peaks, we get PPM4 peaks on the return monitor, and PPM7 peaks result in PPM5 peaks on the return, with greater compression so that it still "feels" much louder, but the audio levels are pretty much under control. Not sure about setting it up to feed two streams though... we'll have to see about that.
All this work resulted in Tim installing shoutcast and an encoder on gateway, and we ran a few tests, and played with the config until I was happy that the capture level wasn't too high (actually, I had to set it very low, because the output levels are set high to allow reasonable throughput to the logging). The upshot is that I can now drive the mixer to PPM7, and while it sounds very loud, it's not too badly distorted, and regular driving to PPM5 sounds reasonably open, with minimum compression... which is how I like my audio. I guess if I had an AGC, I'd set it to drive the compressor a bit harder (maybe up 6dB). This has a knock on effect of improving some tracks that "play" the elbow rather badly. For example "I Know Him So Well", the low levels seem to be pushed down, and then you get a high peak, which goes too high after the compressor has responded to the low level. This causes the track's level to go all over the place. Running it 6dB up seems to resolve this issue, but resulting in an obviously compressed output. Can't win really.
Wednesday was a hard day at work, but at least I can work out for myself why something doesn't work these days. The evening trip to Grantham was uneventful, but received a SMS from Dave inviting me to have a look at some converted PMR kit. How that I've upgraded my ham license, I can now use this kit, so I took up this offer. Ended up drinking whisky until gone midnight...
...making this morning a bit of a slow start. Dragged myself swimming, and did at least 24 lengths, probably more, but I lost count. For some reason, they were piping Radio 1 into the pool area rather than Classic FM. That combined with someone swimming a splashy front crawl in the middle lane (maximum disruption to everyone else), made the morning swim more stressful than the sigmoidoscopy I had last year. Anyway, I now feel rather groggy, sort of mild migrain symptoms, so I'll try to catch a nap later to try to rectify things.
Onwards...
posted by Steve Caddy | 12:05
0 Comments:
Post a Comment
<< Home