Home |
Search |
Today's Posts |
#1
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag?
Which one do you track at? |
#2
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 1/21/2017 11:19 AM, James Price wrote:
Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? Mixed bag. Mostly, I track at 24-bit 44.1 kHz. I rarely track at 96 kHz because I don't have any gear (or ears or sources, for that matter) that needs to work on anything above 20 kHz. I can be more conservative with level and leave more headroom when working at 24-bit than16-bit resolution, so I waste my memory space on headroom that I might use rather than frequency response that I won't use. Some digital signal processors work better at 96 kHz, so if that's how you work, there's an advantage to using that sample rate to start off with. And if your paying client asks for a particular sample rate, you should be prepared to use it. -- "Today's production equipment is IT based and cannot be operated without a passing knowledge of computing, although it seems that it can be operated without a passing knowledge of audio" - John Watkinson Drop by http://mikeriversaudio.wordpress.com now and then |
#3
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Mike Rivers wrote: "And if your paying client asks for a particular sample rate, you
should be prepared to use it. " That is, if said paying client is even remotely as concerned with sampling rate(or other technical matters) as they are with laying down a good effort in as few takes as possible. |
#5
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
James Price wrote:
Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. 2. Your customer demands it. 3. Your converters are imperfect and happen to sound better at the higher rate. And there are three reasons to use the lower sampling rate: 1. You are trying to avoid recording ultrasonics for fear that they will later mix down and form audible distortion products in the audible region. 2. Your customer demands it. 3. Your converters are imperfect and happen to sound better at the lower rate. Which one do you track at? Having done A/B comparisons with my own converters, I think they sound better at the lower rate. This might be because the clock stability is less crucial and it may because of the ultrasonic filtering, or it might be some other reason. But for that reason I will tend to record at 44.1 ksamp/sec. This may not apply to anyone else's situation. Now, that said, I have film customers who want everything at 48 ksamp/sec and I'm happy to oblige them, as well as I am happy to oblige the folks who want recordings at 96. They are paying the bill, if they demand that rate for their own reasons that's fine by me. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#6
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Mike Rivers wrote: "t's no more difficult to record at 2x sample rate, but if that's what
the client asks for, that's what he'll get. " Of course it's no more difficult. My reply was more concerned with client knowledge of the process. Take for example a certain young quartet from Liverpool: Do you think John Lennon would be concerned about sampling rated or bit depths if the Beatles were in ascendency after the year 2000? Not every artist has the same technical prowess as a Don Fagan or Alan Parsons. Most artists today probably are unaware of the technical details, and are more wrapped up in the creative process and just wanting to give it their all. The rest, they leave up to the engineers! |
#7
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
|
#8
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
John Williamson wrote:
"The difference between 16 and 24 bit isn't always obvious on the first generation, but when mixing more than one or two tracks, while applying effects, the difference can become noticeable. Whenever possible, I always record at 24 bit." Absolutely. Use the highest spec one can in production. Akin to giving the dog the 'biggest backyard to play around in' in terms of bit depth and spectrum. Lets the artist be themself during the creative process. |
#9
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Sunday, January 22, 2017 at 7:18:19 AM UTC-6, Scott Dorsey wrote:
James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? |
#10
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Sunday, January 22, 2017 at 11:40:58 AM UTC-6, James Price wrote:
On Sunday, January 22, 2017 at 7:18:19 AM UTC-6, Scott Dorsey wrote: James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Juat about any microphone can capture audio above 22kHz. That audio may be reduced in level, and the response may be dropping at 12dB per octave, but there is some audio there. Meanwhile, I try to record at the sampling rate that will be used in the final product. Recently I did a remastering job, turning a 2-track analog tape from 1980 into a CD. I did the transfer at 96kHz, and did all the editing at that rate (including topping and tailing, level adjustments, and the like). It sounded excellent. Then I converted the sample rate to 44.1, which it needed to be for CD. And all that wonderful sound...went away. The result sounded as grubby as an early-years digital recording. I'd done the SRC in Audition, which had tested well in the Infinite Wave comparison (http://src.infinitewave.ca/). But I wasn't happy. So I took the files to the university, which had iZotope's RX5 on one of their machines -- iZotope's SRC had tested extremely well in the Infinite Wave comparison. Same result. So I redid the transfer, this time at 44.1kHz from the beginning. Luckily, I'd kept detailed notes of all my remastering work, so I could redo it without too much hassle. And I finally had a result I could live with. Bottom line: Sample rate conversion is still a can of worms, and I for one will continue to transfer at the sample rate which will be used for the final product, to avoid having to do SRC in the process. Note that this was an in-the-box remastering job from a 2-track master tape.. If I were doing a multitrack job from scratch, and mixing down in the analog domain, I might be tempted to record at 96k, then (if the result is to be a CD) mix down to a 44.1k 2-track, or 48k if it'a going to be used for video. But I'll avoid SRC as much as possible. YMMV, of course. Peace, Paul |
#11
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
James Price wrote:
On Sunday, January 22, 2017 at 7:18:19 AM UTC-6, Scott Dorsey wrote: James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Yes, and not only that you need a microphone that can capture _significant_ and _useful_ audio above 22 KHz. If you play a fiddle into an SM-57 you'll get get some output above 22 KHz, but it will be nasty junk that isn't adding anything. The wider your bandwidth is, the more noise you have to deal with, and if you don't think the noise at 25 KHz is a problem, wait until you play it back through a dome tweeter and get intermodulation products resulting from it that turn up in the midrange. Wider bandwidth means nonlinear distortion effects become more significant. Of course, you can record at 96 ksamp/sec and then low-pass at 20 KHz anyway, which is what I often do when customers request high sampling rates. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#12
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Sunday, January 22, 2017 at 12:31:55 PM UTC-6, Scott Dorsey wrote:
James Price wrote: On Sunday, January 22, 2017 at 7:18:19 AM UTC-6, Scott Dorsey wrote: James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Yes, and not only that you need a microphone that can capture _significant_ and _useful_ audio above 22 KHz. If you play a fiddle into an SM-57 you'll get get some output above 22 KHz, but it will be nasty junk that isn't adding anything. The wider your bandwidth is, the more noise you have to deal with, and if you don't think the noise at 25 KHz is a problem, wait until you play it back through a dome tweeter and get intermodulation products resulting from it that turn up in the midrange. Wider bandwidth means nonlinear distortion effects become more significant. Of course, you can record at 96 ksamp/sec and then low-pass at 20 KHz anyway, which is what I often do when customers request high sampling rates. If you ultimately low-pass at 20 KHz, what's the advantage (if any) of recording at 24/96 vs. 24/48, other than to appease the client? |
#13
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
James Price wrote:
On Sunday, January 22, 2017 at 12:31:55 PM UTC-6, Scott Dorsey wrote: If you ultimately low-pass at 20 KHz, what's the advantage (if any) of recording at 24/96 vs. 24/48, other than to appease the client? There is none, but the client is happy and it sounds good. In the end, those are really the only two important things ever. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#14
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 1/22/2017 10:48 AM, James Price wrote:
If you ultimately low-pass at 20 KHz, what's the advantage (if any) of recording at 24/96 vs. 24/48, other than to appease the client? There may be none, but don't belittle appeasing the client, unless it's a non-paying client. But if you're working in a DAW, which you almost certainly will be, at least for some part of the project, and you use a processing plug-in that works better at high sample rates, that might be to your advantage. -- "Today's production equipment is IT based and cannot be operated without a passing knowledge of computing, although it seems that it can be operated without a passing knowledge of audio" - John Watkinson Drop by http://mikeriversaudio.wordpress.com now and then |
#15
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
PStamler wrote: "Meanwhile, I try to record at the sampling rate that will be used in the final product.
Recently I did a remastering job, turning a 2-track analog tape from 1980 into a CD. I did the transfer at 96kHz, and did all the editing at that rate (including topping and tailing, level adjustments, and the like). It sounded excellent. Then I converted the sample rate to 44.1, which it needed to be for CD. And all that wonderful sound... went away. The result sounded as grubby as an early-years digital recording. " Anyone wanna venture what happened in his second paragraph? I'm guessing he should have used a sampling rate that was a multiple of that used for the deliverable. I.E. 88.2kHz for a 44.1kHz(CD) product. Is 88.2 avail. as a rate? |
#16
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
wrote:
PStamler wrote: "Meanwhile, I try to record at the sampling rate that will be used in the final product. Recently I did a remastering job, turning a 2-track analog tape from 1980 into a CD. I did the transfer at 96kHz, and did all the editing at that rate (including topping and tailing, level adjustments, and the like). It sounded excellent. Then I converted the sample rate to 44.1, which it needed to be for CD. And all that wonderful sound... went away. The result sounded as grubby as an early-years digital recording. " Anyone wanna venture what happened in his second paragraph? I'm guessing he should have used a sampling rate that was a multiple of that used for the deliverable. I.E. 88.2kHz for a 44.1kHz(CD) product. Is 88.2 avail. as a rate? Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#17
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 22/01/2017 5:53 PM, Mike Rivers wrote:
.. It's no more difficult to record at 2x sample rate, but if that's what the client asks for, that's what he'll get. Getting good recordings is his job at whatever sample rate. I'll concede that some of them don't know what sample rate means, but they may have read that 96 kHz is good, or have been asked by the people who are paying the bills to get 96 kHz recordings because it's company policy, or the recording you're making will be part of a project with recordings coming in from other places, and if they're all at the same sample rate, it's easier to combine them into a CD than if the recordings were at multiple sample rates. Higher sample rates can be demanded by those with little understanding, such as with 'loudness'. When we were using tape, doubling the tape speed meant using twice as much tape, which could make a big dent in the project's budget. Accommodating files at 2x sample rate on a disk drive may not add any cost at all, or maybe only a few bucks. On the other hand, when you're burning up tape twice as fast, you might not want to keep all of those takes that might have a better word in them than your primary take. That could save some time if you're not afraid to make a decision. Surely no reason for anybody to ever track with less than 24-bit bitdepth ? The potential 'cost' of higher sample rates (why stop at 88 or 96, what about 192 !!!) is degradation of performance of a DAW wrt CPU loading, or capability of the transfer method between A-D and DAW ... geoff |
#18
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
|
#19
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
PStamler wrote:
On Sunday, January 22, 2017 at 11:40:58 AM UTC-6, James Price wrote: snip Bottom line: Sample rate conversion is still a can of worms, and I for one will continue to transfer at the sample rate which will be used for the final product, to avoid having to do SRC in the process. snip You'd need two full setups of course, but it seems like simply transferring in real time via analog would be preferrable to SRC. YMMV, of course. Peace, Paul -- Les Cargill |
#20
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 1/23/2017 4:47 AM, Les Cargill wrote:
You'd need two full setups of course, but it seems like simply transferring in real time via analog would be preferrable to SRC. I agree with Paul Stamler that the best approach to sample rate conversion is to avoid it. It seems that there are good SRC processes and not-so-good ones, just like there are good and not-so-good D/A and A/D converters, and good sample rate conversion requires a lot of computer horsepower. You have to, in essence, reconstruct the original waveform of the source, then re-sample the (virtually) reconstructed waveform at your desired rate. If the reconstruction isn't accurate, you'll end up re-sampling interpolated values, and that's what doesn't sound right. A while back, someone, maybe it was Mytek, made a converter that could use separate clocks for the D/A and A/D sections, so you could just pipe the output into the input and convert the sample rate in a single hardware box. It sounded better than any of the software SRC programs of its time. But then, Mytek's converters were among the best sounding ones in the industry, so they had a good start. -- "Today's production equipment is IT based and cannot be operated without a passing knowledge of computing, although it seems that it can be operated without a passing knowledge of audio" - John Watkinson Drop by http://mikeriversaudio.wordpress.com now and then |
#21
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Sunday, January 22, 2017 at 12:54:22 PM UTC-6, Scott Dorsey wrote:
James Price wrote: On Sunday, January 22, 2017 at 12:31:55 PM UTC-6, Scott Dorsey wrote: If you ultimately low-pass at 20 KHz, what's the advantage (if any) of recording at 24/96 vs. 24/48, other than to appease the client? There is none, but the client is happy and it sounds good. In the end, those are really the only two important things ever. In Mastering Audio, Bob Katz indicated that one benefit of higher sampling rates (eg. 96 kHz) would be moving unwanted converter noise above the audible frequency range, some of which is filtered out upon downsampling. The main idea being that moving the filter cutoff to 48 kHz (for 96 kHz SR) relaxes the filter requirement and makes it easier to design filters with less ripple in the passband and less phase shift near the upper frequency limit. Any thoughts? |
#22
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
James Price wrote:
On Sunday, January 22, 2017 at 12:54:22 PM UTC-6, Scott Dorsey wrote: James Price wrote: On Sunday, January 22, 2017 at 12:31:55 PM UTC-6, Scott Dorsey wrote: If you ultimately low-pass at 20 KHz, what's the advantage (if any) of recording at 24/96 vs. 24/48, other than to appease the client? There is none, but the client is happy and it sounds good. In the end, those are really the only two important things ever. In Mastering Audio, Bob Katz indicated that one benefit of higher sampling rates (eg. 96 kHz) would be moving unwanted converter noise above the audible frequency range, some of which is filtered out upon downsampling. The main idea being that moving the filter cutoff to 48 kHz (for 96 kHz SR) relaxes the filter requirement and makes it easier to design filters with less ripple in the passband and less phase shift near the upper frequency limit. This was absolutely critical back in 1985, and it was why you saw higher rates used on things like the Mitsubishi recorders. But then we got oversampling everywhere which did the same thing without having to record at the higher rate. And then the whole world changed with sigma-delta converters where the noise issues are totally different. Any thoughts? Thank God we don't deal with ladder converters that are barely able to keep up with the sample rate and drift all over the place today. The sigma-delta technology has made getting real linearity across a wide range far less expensive and made brickwall filters a thing of the past. It was a real and actual revolution. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#23
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Scott Dorsey wrote:
Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott What is the good SRC? Paul Stamler mentioned being dissatisfied with iZotope, which I thought was supposed to be one of the good ones. I've also heard mention that Weiss Saracon is good, but it's far from free. |
#24
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Mon, 23 Jan 2017 15:29:07 -0600,
lid (Tatonik) wrote: Scott Dorsey wrote: Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott What is the good SRC? Paul Stamler mentioned being dissatisfied with iZotope, which I thought was supposed to be one of the good ones. I've also heard mention that Weiss Saracon is good, but it's far from free. Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ d |
#25
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Monday, January 23, 2017 at 3:41:11 PM UTC-6, Don Pearce wrote:
On Mon, 23 Jan 2017 15:29:07 -0600, lid (Tatonik) wrote: Scott Dorsey wrote: Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott What is the good SRC? Paul Stamler mentioned being dissatisfied with iZotope, which I thought was supposed to be one of the good ones. I've also heard mention that Weiss Saracon is good, but it's far from free. Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ d That was my resource in choosing the SRCs from Audition and iZotope. Both did very well in the site's test, but I didn't like the results of either as well as I did my redub at 44.1, with no SRC. How come? I don't know. Peace, Paul |
#26
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Don Pearce wrote:
On Mon, 23 Jan 2017 15:29:07 -0600, (Tatonik) wrote: Scott Dorsey wrote: Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott What is the good SRC? Paul Stamler mentioned being dissatisfied with iZotope, which I thought was supposed to be one of the good ones. I've also heard mention that Weiss Saracon is good, but it's far from free. Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ Wow, this is an interesting site. It doesn't look like the detail is enough so you can tell the differences between the best converters, but it sure makes it obvious what the differences between the best and not-so-best are. I wish this had some of the hardware sample rate converters on there, just to show how bad most of them are. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#27
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Monday, January 23, 2017 at 3:41:11 PM UTC-6, Don Pearce wrote:
Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ The graphs for Adobe Audition 5.5 are pretty impressive. |
#28
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
So what should we see with those graphs?
The fewer cross-hatch lines the better? |
#29
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Tuesday, January 24, 2017 at 1:36:08 PM UTC-5, wrote:
So what should we see with those graphs? The fewer cross-hatch lines the better? Too many bean counters. Forget graphs, use God given ears, and you too, will receive compliments like this.... "Ive read for years (on BSN) that the double tracked vocal was missing from the multi track, and was dubbed in mono onto the final master. Your version blows that theory out of the water!" Jack |
#30
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 23/01/2017 7:31 a.m., PStamler wrote:
On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Juat about any microphone can capture audio above 22kHz. That audio may be reduced in level, and the response may be dropping at 12dB per octave, but there is some audio there. All tracking should be done at 3kHz. geoff |
#31
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Don Pearce wrote:
On Mon, 23 Jan 2017 15:29:07 -0600, lid (Tatonik) wrote: Scott Dorsey wrote: Sadly, most SRC programs won't filter-and-decimate when they encounter a multiple, they just use the same algorithm they use for non-multiples. So having sample rates that are multiples doesn't actually help you. Good SRC has become very good these days, but bad SRC remains very very bad. But there is good free SRC available if you don't need it to work in realtime. --scott What is the good SRC? Paul Stamler mentioned being dissatisfied with iZotope, which I thought was supposed to be one of the good ones. I've also heard mention that Weiss Saracon is good, but it's far from free. Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ d Audacity comes out well in that. Interesting. --- Les Cargill |
#32
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On 24/01/2017 1:32 AM, Mike Rivers wrote:
On 1/23/2017 4:47 AM, Les Cargill wrote: You'd need two full setups of course, but it seems like simply transferring in real time via analog would be preferrable to SRC. I agree with Paul Stamler that the best approach to sample rate conversion is to avoid it. It seems that there are good SRC processes and not-so-good ones, just like there are good and not-so-good D/A and A/D converters, and good sample rate conversion requires a lot of computer horsepower. Which any computer made in the last decade has plenty of for any audio process. You have to, in essence, reconstruct the original waveform of the source, then re-sample the (virtually) reconstructed waveform at your desired rate. If the reconstruction isn't accurate, you'll end up re-sampling interpolated values, and that's what doesn't sound right. And the only reason for that is bad programming. Which sadly does abound of course. A while back, someone, maybe it was Mytek, made a converter that could use separate clocks for the D/A and A/D sections, so you could just pipe the output into the input and convert the sample rate in a single hardware box. It sounded better than any of the software SRC programs of its time. But then, Mytek's converters were among the best sounding ones in the industry, so they had a good start. Yep, good DA-AD is going to beat bad software, and vice versa. Trevor. |
#33
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Sunday, January 22, 2017 at 1:31:10 PM UTC-5, PStamler wrote:
On Sunday, January 22, 2017 at 11:40:58 AM UTC-6, James Price wrote: On Sunday, January 22, 2017 at 7:18:19 AM UTC-6, Scott Dorsey wrote: James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? If the converters are perfect, the only thing the higher sampling rate buys you is added bandwidth. So, there are three reasons to use the higher sampling rate: 1. You are trying to record ultrasonics for later analysis that may make them audible or create audible effects from them. Say you are recording bird song for analysis or machine vibrations or you're making dubs of LPs that will later go through noise reduction which will benefit from the ultrasonic content for pop detection. On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Juat about any microphone can capture audio above 22kHz. That audio may be reduced in level, and the response may be dropping at 12dB per octave, but there is some audio there. Meanwhile, I try to record at the sampling rate that will be used in the final product. Recently I did a remastering job, turning a 2-track analog tape from 1980 into a CD. I did the transfer at 96kHz, and did all the editing at that rate (including topping and tailing, level adjustments, and the like). It sounded excellent. Then I converted the sample rate to 44.1, which it needed to be for CD. And all that wonderful sound...went away. The result sounded as grubby as an early-years digital recording. I'd done the SRC in Audition, which had tested well in the Infinite Wave comparison (http://src.infinitewave.ca/). But I wasn't happy. So I took the files to the university, which had iZotope's RX5 on one of their machines -- iZotope's SRC had tested extremely well in the Infinite Wave comparison. Same result. So I redid the transfer, this time at 44.1kHz from the beginning. Luckily, I'd kept detailed notes of all my remastering work, so I could redo it without too much hassle. And I finally had a result I could live with. Bottom line: Sample rate conversion is still a can of worms, and I for one will continue to transfer at the sample rate which will be used for the final product, to avoid having to do SRC in the process. Note that this was an in-the-box remastering job from a 2-track master tape. If I were doing a multitrack job from scratch, and mixing down in the analog domain, I might be tempted to record at 96k, then (if the result is to be a CD) mix down to a 44.1k 2-track, or 48k if it'a going to be used for video. But I'll avoid SRC as much as possible. YMMV, of course. Peace, Paul Still can't understand how you can remaster a master. Digitally enhance, yes, but no remastering. Allow me to explain. This Asia '80's hit was remixed from (3) stereo .ogg files. Obviously, more than (6) tracks were actually used, someone premixed the multiple tracks to minimize file trading size... http://www.angelfire.com/empire/abps...atofmoment.mp3 Note: No 3kHz were harmed during the remixing of this fine tune. Jack |
#34
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
Les Cargill wrote:
Don Pearce wrote: On Mon, 23 Jan 2017 15:29:07 -0600, Have a look here. Someone has done a lot of work comparing the quality of various SRCs. http://src.infinitewave.ca/ Audacity comes out well in that. Interesting. Not surprising, seeing that it's basically designed by mathematicians and dsp people, and because it does the conversion as batch it has all the time in the world and doesn't need to take any shortcuts. The Audacity UI is just horrible, but the compute engine is hard to beat inside. Contrast this with say the old Sonic Studios systems that tried to do the conversion in realtime on a 68k. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#35
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
geoff wrote:
On 23/01/2017 7:31 a.m., PStamler wrote: On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Juat about any microphone can capture audio above 22kHz. That audio may be reduced in level, and the response may be dropping at 12dB per octave, but there is some audio there. All tracking should be done at 3kHz. geoff Tree-fiddy. -- Les Cargill |
#36
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Wednesday, January 25, 2017 at 7:13:14 PM UTC-5, Les Cargill wrote:
geoff wrote: On 23/01/2017 7:31 a.m., PStamler wrote: On a side note, is the ability to record at 24/96 limited by the frequency response of the microphone? Wouldn't you need a mic that can capture audio above 22kHz? Juat about any microphone can capture audio above 22kHz. That audio may be reduced in level, and the response may be dropping at 12dB per octave, but there is some audio there. All tracking should be done at 3kHz. geoff Tree-fiddy. You might tell Mr. G', I've been receiving some nice site compliments. No, not just on audio content, but I explain the REAL WORLD of (re)mastering. Anyway, don't hold Mr. G' up, I'm sure he has to rush to buy some new headphones so he can appreciate a song with, ahem, great (re)mastering. Jack -- Les Cargill |
#37
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Saturday, January 21, 2017 at 2:19:43 PM UTC-5, James Price wrote:
Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? Which one do you track at? the reason to do all first / best need quality recording of raw tracks w/ at least 24/48 or 24/96 is not just individual track dynamic range - which is way beyond the dynamics of human music. The main reason in the digital age to use 24/48 at least (and 24/96 is ez to achieve these days in A/D/A and DSP systems and also allows very gentle slop anti-aliasing filters that start above human hearing to be used, which essentially have no phase effect on the 20-20 khz audio range music signal) is more the 24 bit over 16 bit part than the 48 versus 96 sample rate. When all the multitrack mix down and EQ is done digitally with DSP (in the box or with outboard digital gear) as opposed to being processed with outboard analog gear (like neve and or SSL quote current and or vintage super analog stuff) hooked to the multichannel A/D/A systems - is due to DSP math processing and acuumulated and compounded round off errors. All the DAW mixer and EQ engines these days (pro tools, cubase, sonar, etc) are 64 bit - and when processing 24 bit wav files -- the round off error all stay way above impacting the 24 accuracies of the individual track wav files...... i have listened to 16 and 24 bit wav files processed with 40 bit and 64 bit DSP - the 16 with 40 bit is horrid, the 24 bit is better but the 24 bit processed with DAWs and digital outboard gear using 64 bit math - are VERY audibly better overall, and ez to hear the difference even when dithered down to 16/44.1 CD and or compressed with a high rate MP3. |
#38
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
On Thursday, November 2, 2017 at 4:12:06 PM UTC-4, wrote:
On Saturday, January 21, 2017 at 2:19:43 PM UTC-5, James Price wrote: Do you think it's the norm for most studios to track at 24/96 or 24/48? Or is it a mixed bag? Which one do you track at? the reason to do all first / best need quality recording of raw tracks w/ at least 24/48 or 24/96 is not just individual track dynamic range - which is way beyond the dynamics of human music. The main reason in the digital age to use 24/48 at least (and 24/96 is ez to achieve these days in A/D/A and DSP systems and also allows very gentle slop anti-aliasing filters that start above human hearing to be used, which essentially have no phase effect on the 20-20 khz audio range music signal) is more the 24 bit over 16 bit part than the 48 versus 96 sample rate. When all the multitrack mix down and EQ is done digitally with DSP (in the box or with outboard digital gear) as opposed to being processed with outboard analog gear (like neve and or SSL quote current and or vintage super analog stuff) hooked to the multichannel A/D/A systems - is due to DSP math processing and acuumulated and compounded round off errors. All the DAW mixer and EQ engines these days (pro tools, cubase, sonar, etc) are 64 bit - and when processing 24 bit wav files -- the round off error all stay way above impacting the 24 accuracies of the individual track wav files...... i have listened to 16 and 24 bit wav files processed with 40 bit and 64 bit DSP - the 16 with 40 bit is horrid, the 24 bit is better but the 24 bit processed with DAWs and digital outboard gear using 64 bit math - are VERY audibly better overall, and ez to hear the difference even when dithered down to 16/44.1 CD and or compressed with a high rate MP3. Regardless of technology and even the highest bit and sample rates, man will always be the weakest link when it comes to music. Jack |
#39
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
i have listened to 16 and 24 bit wav files processed with 40 bit and 64 bit DSP - the 16 with 40 bit is horrid, the 24 bit is better but the 24 bit processed with DAWs and digital outboard gear using 64 bit math - are VERY audibly better overall, and ez to hear the difference even when dithered down to 16/44.1 CD and or compressed with a high rate MP3.
..... can you make a quantitative measurement to demonstrate this. m |
#40
Posted to rec.audio.pro
|
|||
|
|||
Tracking at 24/96 vs 24/48
wrote:
i have listened to 16 and 24 bit wav files processed with 40 bit and 64 bit= DSP - the 16 with 40 bit is horrid, the 24 bit is better but the 24 bit p= rocessed with DAWs and digital outboard gear using 64 bit math - are VERY a= udibly better overall, and ez to hear the difference even when dithered dow= n to 16/44.1 CD and or compressed with a high rate MP3. .... can you make a quantitative measurement to demonstrate this. Probably not anymore. I think you'll be hard-pressed to find anything still doing fixed-point math for audio applications any longer. Maybe some cheaper standalone gear. But then, I tend to avoid doing any processing in the digital domain at all. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
Reply |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
drum tracking | Pro Audio | |||
Do You Need Millennium Plus GPS Car Tracking? | Car Audio | |||
Tracking in PT / Mix in DP5? | Pro Audio | |||
Tracking with 2 Firepods | Pro Audio | |||
Tracking Headphones | Pro Audio |