Audio Quality
Moderators: snarkout, Patrick, dann
Audio Quality
I was just wondering why Dave Yates' "carcast" has better sound quality than yours? I can hear his voice loud and clear, but your individual volumes are all over the place. Your guest is usually booming, Dann is normal, and the rest of you are all pretty quiet. This is my only criticism of your show. Great show!
As a podcaster myself, it's relatively easy to record and edit a show, especially a one-man show, like the one I do or Dave's show, where you are in total control of all the elements. It is far, far more difficult to put on a 2+ hour show with four hosts and 1-2 guests broadcast live over the internet like the TLLTS guys do and have done for more than 3 years. I think what they do is amazing considering the budget (close to zero, I imagine) and the fact they are spread out all over the place.
Chess Griffin
Most of the podcast are digital quality while TLLTS do a lot of changing from analog to digital quality. Actually that might be fine except is analog over the telephone which might not always be a good thing.
Also other things such as bandwith latency, skype interference and other factors might make the quality vary from show to show. By the speed the show gets posted I doubt it gets that much post-editing.
Also other things such as bandwith latency, skype interference and other factors might make the quality vary from show to show. By the speed the show gets posted I doubt it gets that much post-editing.
Alexandro COLORADO
believe me, I would love nothing more than to fix our audio issues. Like previously said, it is so damn difficult to get everyone's sound level the same. We are limited on resources and everyone eventuallly comes into a single channel. So until we can get a seperate channel for each remote source so we can adjust the levels there is practically no way we are going to get the quality we desire.
We are hopefully making some changes, but it's very difficult.
We are hopefully making some changes, but it's very difficult.
I really didn't mean it as a slam. I was just curious about it. Have you ever tried a compressor? A compressor levels all incoming signals. Even a single channel will be helped by one. I know Audacity ships with one, but I must admit I have never tried it. Being a musician, I have worked with compressors and I know good quality ones have several adjustments. I think rezound also has a compressor built in.
-
Wally Balljacker
- Posts: 1227
- Joined: Fri Jul 29, 2005 3:32 am
- Location: University of Massachusetts - Lowell
- Contact:
It's funny, I have a compression pedal for my guitar, but never actually understood how it worked until we got a recording machine and I started reading up on it.
Compressors actually do just that, and compress any loud signals, cutting them down to a softer level. Of course, most use that compression and then increase the overall sound, but the compression itself merely cuts down the loud signals, making them softer, generally after a short "attack" time where sounds are allowed to become louder.
The problem with having so many signals probably wouldn't be helped by a compressor. Until the other signals are brought to the same level, the compressor would merely cut the loudest portions of the loudest signal, and leave the rest alone. My band tries to do live recordings, and we routinely get our asses kicked by trying to even out the levels for the vocals, guitar, bass, and drums. We can only record to two channels, which just carries the stereo signal, and can't adjust the volume of the individual instruments later.
I imagine that's the same problem the tech show has, and the only thing we've been able to do is just really try hard to get the levels in good order by testing the recording with a couple of songs before we start a gig. Once it's on the digital recorder, though, there's no going back and increasing or decreasing the volume of any one instrument. No amount of compression has ever helped, either.
[EDIT: An addition. One might think you could conceivably use a compressor to simply compress the volumes down to almost nothing, squeezing the louder signals down to the same level as the softer signals, and then finally raise the volume of that ridiculously compressed signal. You could do that, but you would also increase the noise—which will always be considerable with VOIP or any other telephone—to possibly unbearable levels, not to mention giving the show a very artificial sound.
I think there are really only two methods that can help, 1) recording to separate channels, or 2) following a significantly long preparation routine before the beginning of the show along the lines of: test record, relevel, repeat. This would also have to be done with the guest as well to get his level correct. That's just another burden on top of what is already a tough show to manage.
So, who's going to donate the multichannel recording device? Barring that, since the guys only have one channel to work with anyway, there's no reason why one of us couldn't test the compress-it-to-oblivion technique ourselves and share our results with the group, or volunteer to do it for the show.]
Compressors actually do just that, and compress any loud signals, cutting them down to a softer level. Of course, most use that compression and then increase the overall sound, but the compression itself merely cuts down the loud signals, making them softer, generally after a short "attack" time where sounds are allowed to become louder.
The problem with having so many signals probably wouldn't be helped by a compressor. Until the other signals are brought to the same level, the compressor would merely cut the loudest portions of the loudest signal, and leave the rest alone. My band tries to do live recordings, and we routinely get our asses kicked by trying to even out the levels for the vocals, guitar, bass, and drums. We can only record to two channels, which just carries the stereo signal, and can't adjust the volume of the individual instruments later.
I imagine that's the same problem the tech show has, and the only thing we've been able to do is just really try hard to get the levels in good order by testing the recording with a couple of songs before we start a gig. Once it's on the digital recorder, though, there's no going back and increasing or decreasing the volume of any one instrument. No amount of compression has ever helped, either.
[EDIT: An addition. One might think you could conceivably use a compressor to simply compress the volumes down to almost nothing, squeezing the louder signals down to the same level as the softer signals, and then finally raise the volume of that ridiculously compressed signal. You could do that, but you would also increase the noise—which will always be considerable with VOIP or any other telephone—to possibly unbearable levels, not to mention giving the show a very artificial sound.
I think there are really only two methods that can help, 1) recording to separate channels, or 2) following a significantly long preparation routine before the beginning of the show along the lines of: test record, relevel, repeat. This would also have to be done with the guest as well to get his level correct. That's just another burden on top of what is already a tough show to manage.
So, who's going to donate the multichannel recording device? Barring that, since the guys only have one channel to work with anyway, there's no reason why one of us couldn't test the compress-it-to-oblivion technique ourselves and share our results with the group, or volunteer to do it for the show.]
Vim is beautiful
Vogateer, you almost hit it on the head. But, in addition to a multi-track recorder we also need multiple sources. Have a multi-track recorder and a single or two sources does next to nothing.
Up until last week we had 2 channels. One was my mic into the mixer and the other was the asterisk conference/skype channel. The latter is an all or nothing bag. We can adjust some of the volume on the other end since the other three guys are all going into their own mixer. They can adjust their individual volumes to some extent but it comes to me as one stream.
I can adjust my volume to match their single stream but it's hard to account for any further fluctuations and/or quality issues as time goes on.
Now throw in a third, fourth or more stream coming into the asterisk conference/skype channel and the problem further degrades. When the guest's volume is such that their stream is coming to me under -15 db and everyone else is above -8 db there's almost no way to correct. We can lower the other three guys on their end to hopefully match the guest and then bring the whole channel up, but more often than not, when there is an issue, we are already getting close to max volume on that channel.
We just put in a third source and hopefully that will help correct some of the problems.
In addition to having multiple sources in we also need multiple channels in. So that having two asterisk clients running the first needs to hear all the channels execpt their own and the second needs to hear all channels but their own. On my end I believe we are out of channels to do that now. We can support 2 external sources like this.
Up until last week we had 2 channels. One was my mic into the mixer and the other was the asterisk conference/skype channel. The latter is an all or nothing bag. We can adjust some of the volume on the other end since the other three guys are all going into their own mixer. They can adjust their individual volumes to some extent but it comes to me as one stream.
I can adjust my volume to match their single stream but it's hard to account for any further fluctuations and/or quality issues as time goes on.
Now throw in a third, fourth or more stream coming into the asterisk conference/skype channel and the problem further degrades. When the guest's volume is such that their stream is coming to me under -15 db and everyone else is above -8 db there's almost no way to correct. We can lower the other three guys on their end to hopefully match the guest and then bring the whole channel up, but more often than not, when there is an issue, we are already getting close to max volume on that channel.
We just put in a third source and hopefully that will help correct some of the problems.
In addition to having multiple sources in we also need multiple channels in. So that having two asterisk clients running the first needs to hear all the channels execpt their own and the second needs to hear all channels but their own. On my end I believe we are out of channels to do that now. We can support 2 external sources like this.
While the audio quality continues to be a thorn in our side; I'm not taking what you said as a slam.toddr wrote:I really didn't mean it as a slam. I was just curious about it. Have you ever tried a compressor? A compressor levels all incoming signals. Even a single channel will be helped by one. I know Audacity ships with one, but I must admit I have never tried it. Being a musician, I have worked with compressors and I know good quality ones have several adjustments. I think rezound also has a compressor built in.
We do have compressor and I have tried post production. Most of the time post production, while it may help a bit, tends to bring to much noise to the file. Plus, post production on a 2+ hour show take a hell of a lot of time. I usually have the file up within an hour after the show with some post (mostly trying to remove silence these days). Any more and expect the file the next night or weekend.
I just tested the latest mp3 file in Audacity, and had practically no success using compression. I compressed the Hell out of it, too. Chopped things off at a few places between -29 and -58, compressed multiple times in the same area, and still had pretty severe variance in the voices.
The only thing I could do that seemed to work was by far the most time consuming. I used the enveloping feature to squeeze the volume in some places and expand them in others. It works impressively well, and would probably take well over 20 hours to do that to the entire two hour show, and that's being optimistic.
Obviously I haven't had much time to try things out, but the results I had were pretty abysmal. Basically, if you only have one channel and you don't start off with a good sound going into post production, you're already screwed. I didn't think things through enough to realize that both Jeremy Allison and the rest of the crew were coming in on the same channel. At least that's the way I'm assuming things work. That makes the multi-track recording pretty much useless. Dann might be able to record himself separately from the rest of the voices, but then he has to decide to level himself with the guest or with the rest of the crew.
Once again, the only solution there I see is ridiculously time consuming, and would require the guest, Dann, and the other guys all recording their own voices on their own computers, and then sending them to one person who edits them together later. That's just brutal. The bandwidth alone for wav files would be nasty, and compressing to mp3 followed by effects and manipulations just further degrades the sound.
As far as I can tell, recording a live show is pure evil. No one person has good control of the sound with the current setup, and the limitations of Asterisk and Skype make multichannel recording unrealistic.
The only thing I could do that seemed to work was by far the most time consuming. I used the enveloping feature to squeeze the volume in some places and expand them in others. It works impressively well, and would probably take well over 20 hours to do that to the entire two hour show, and that's being optimistic.
Obviously I haven't had much time to try things out, but the results I had were pretty abysmal. Basically, if you only have one channel and you don't start off with a good sound going into post production, you're already screwed. I didn't think things through enough to realize that both Jeremy Allison and the rest of the crew were coming in on the same channel. At least that's the way I'm assuming things work. That makes the multi-track recording pretty much useless. Dann might be able to record himself separately from the rest of the voices, but then he has to decide to level himself with the guest or with the rest of the crew.
Once again, the only solution there I see is ridiculously time consuming, and would require the guest, Dann, and the other guys all recording their own voices on their own computers, and then sending them to one person who edits them together later. That's just brutal. The bandwidth alone for wav files would be nasty, and compressing to mp3 followed by effects and manipulations just further degrades the sound.
As far as I can tell, recording a live show is pure evil. No one person has good control of the sound with the current setup, and the limitations of Asterisk and Skype make multichannel recording unrealistic.
Vim is beautiful
I've been testing with Ardour and the new mixer. I can record up to 4 channels simultaneously. I can record the PA crew locally and have Dan and the guest on the Asterisk line coming in. At the very least I could bump up the level of the Asterisk channel after the initial recording. We still need to make sure Dan is the same volume as the guest. I did a test one night with me and Allan at my place and Dan was way louder than the guest which I couldn't do anything about it after the fact. Once the mancave is finished I may offer to host Allan and Linc and start doing some of the recordings with Ardour.
Last edited by Patrick on Thu Jan 18, 2007 2:29 pm, edited 1 time in total.
Ego contemno licentia
-
Wally Balljacker
- Posts: 1227
- Joined: Fri Jul 29, 2005 3:32 am
- Location: University of Massachusetts - Lowell
- Contact:
I do like the idea of each recording locally, but I still worry about the time of post production. As far as I can see, you'd either have to not record the PA crew on Dann's end and risk not getting them at all if something goes wrong, edit them out of Dann's recording and lose a ton of time trying to do so—unless I'm getting confused, the PA crew and the guest are on the same channel, so you can't just drop that channel and add the PA crew's recording without losing the guest—or try to get them perfectly synced so that they overlay Dann's recording exactly. Does audio recording experience something similar to frame drops in video? That would be a deal breaker if it's so.
You can try to have each end of the recording clap their hands at the same time to get in sync, but then the lag would probably make that a royal pain. I don't see a silver bullet, but multi-track recording in Ardour would probably be the best improvement without taking a ton of time.
You can try to have each end of the recording clap their hands at the same time to get in sync, but then the lag would probably make that a royal pain. I don't see a silver bullet, but multi-track recording in Ardour would probably be the best improvement without taking a ton of time.
Vim is beautiful
Well the card we want to go with for 8 track recording in Linux is the M-Audio 1010 with the breakout box:
This sucker goes for $399-$499 new. Cheaper used. Before doing a donation drive I think we should do more testing with the M-audio Delta44 I have. We should try to do a test with 3 concurrent voip connections on separate channels and see if there's any bleedover or echo. If the test is successful we may do a pledge drive for the new equipment.
This sucker goes for $399-$499 new. Cheaper used. Before doing a donation drive I think we should do more testing with the M-audio Delta44 I have. We should try to do a test with 3 concurrent voip connections on separate channels and see if there's any bleedover or echo. If the test is successful we may do a pledge drive for the new equipment.
Ego contemno licentia