are you saying no recorder is needed?

[killfilm]

since the redone will support 24bit depth and 48khz audio, is their any need for an external recorder?

[I Bloom]

since the redone will support 24bit depth and 48khz audio, is their any need for an external recorder?

Well you still need to get the audio into the camera, which means tethering it or mounting wireless receivers. The sound department might decide to do seperate sound for various reasons, especially since you can sync timecode with RED. Some sound guys swear by things like pre-amps and stuff, and in RED we don't know yet what we are getting.

But the short answer to your question is: Most people will not use an external recorder, because the sound options on camera look like they are going to be pretty awesome and it can greatly simplify post.

Plus I think its 96Khz. Even better right.

IBloom

[killfilm]

24bit, 96khz would be freakin awesome.

[Kevin Halverson]

The spec does mention 24 bit sample depth, however, the sample rate is 48 kS/s. Of the two, I think the greater word length will be much more valuable than the increased bandwidth that a higher sample rate would provide.

[killfilm]

good point, but now i'm thinking twice about purchasing a seperate recorder,
is this justified?

[Kevin Halverson]

From what I understand, audio won't be enabled in the first feature set, so if you getting a camera early, then you will need another solution. I would suggest that you rent a field recorder and do a parallel output from your field mixer to both the field recorder and the RED. Then you can make a decision later. I suspect that once implemented, the RED will be a perfectly capable audio device, but the cabling requirements might still make a field recorder a better option for some situations.

[Michael Hastings]

The spec does mention 24 bit sample depth, however, the sample rate is 48 kS/s. Of the two, I think the greater word length will be much more valuable than the increased bandwidth that a higher sample rate would provide.

48KHz should give frequency response to 24,000 cycles - higher than virtually any human. Unless you are producing for the burgeoning dog and cat markets. As an underwater shooter, I am looking for a 300KHz recording solution to cover the Whales and dolphins market. Since I won't be doing a lot of added effects to the sounds I think I can get by with 300KHz/16Bit. :sarcasm:

Remember we are all going to be underwater once the ice caps melt.


Humans 20- 20,000

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. (My rock & roll bruised eardrums can't hear much past 10K)

Cats 100- 32,000
Dogs 40- 46,000
Horses 31- 40,000
Elephants 16- 12,000
Cattle 16- 40,000
Bats 1,000- 150,000
Grasshoppers and locusts 100- 50,000
Rodents 1,000- 100,000
Whales and dolphins 70- 150,000
Seals and sea lions 200- 55,000

[Jarred Land]

damn we forgot about the grasshoppers.

[Shawn Nelson]

I am appalled at this blatant grasshopper-phobic bigotry.

[laguun]

damn we forgot about the grasshoppers.

happens all the time!

[Vincent Rice]

48KHz should give frequency response to 24,000 cycles - higher than virtually any human. Unless you are producing for the burgeoning dog and cat markets. As an underwater shooter, I am looking for a 300KHz recording solution to cover the Whales and dolphins market. Since I won't be doing a lot of added effects to the sounds I think I can get by with 300KHz/16Bit. :sarcasm:

Remember we are all going to be underwater once the ice caps melt.


Humans 20- 20,000

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. (My rock & roll bruised eardrums can't hear much past 10K)

Cats 100- 32,000
Dogs 40- 46,000
Horses 31- 40,000
Elephants 16- 12,000
Cattle 16- 40,000
Bats 1,000- 150,000
Grasshoppers and locusts 100- 50,000
Rodents 1,000- 100,000
Whales and dolphins 70- 150,000
Seals and sea lions 200- 55,000

Studios don't record at 96K for the frequency response. Its all about transients.

[G.A. Kokes]

Studios don't record at 96K for the frequency response. Its all about transients.

Exactly.

Harmonics of these frequencies are audible to many. 96k would be a very welcome feature. I believe some of the latest audio tracks on blu ray are encoded 24bit 96k.(I may be wrong. Things are changing very fast in that field)

DVD-Audio goes even higher...192k at 24bit on some 2 channel recordings. In this case I have found that only audio previously recorded in analog IE infinite sampling, benifits. Listening to some recordings, you would swear you were hearing the original master tapes. I am not suggesting 192k on RED.

At 24bit 96k, your mic, mixer, sound guy etc better be up to snuff or you won't benefit.

G

[GlennChan]

48KHz should give frequency response to 24,000 cycles - higher than virtually any human.
48khz sampling doesn't ensure you'll get 48khz performance.

In practical implementations, you need to filter off frequencies higher than 48khz... otherwise aliasing will occur. When you do this filtering, you either get a reduction in frequency response. Or, you oversample (e.g. at 96khz, at 192khz, etc.) and apply a different sort of filtering... the side effect of which are ringing artifacts (but you can get nice frequency response).

The offshoot of that is that 48khz implementations differ in performance (e.g. depending on what filtering they use, whether there is oversampling, etc.).

2- A practical thing to do would be to compare the actual performance of 48khz and 96khz equipment that is being sold. An example would be:

http://prorec.com/Articles/tabid/109/EntryId/158/Default.aspx
(scroll down...)

3- For recording dialogue, your #1 problem is usually too much background noise. How many low-budget films have you heard with bad audio? 24khz 96/192khz would not help these films.

4- For bigger budget productions, you'd likely record onto a separate recorder anyways. You would get a lot more tracks... and high quality preamps+A/Ds.

[Michele Gavazzeni]

The sound department might decide to do seperate sound for various reasons, especially since you can sync timecode with RED.

what is an easy way to achive timecode sync for external sound recording?

[PenGun]

48KHz should give frequency response to 24,000 cycles - higher than virtually any human. Unless you are producing for the burgeoning dog and cat markets. As an underwater shooter, I am looking for a 300KHz recording solution to cover the Whales and dolphins market. Since I won't be doing a lot of added effects to the sounds I think I can get by with 300KHz/16Bit. :sarcasm:

Remember we are all going to be underwater once the ice caps melt.


Humans 20- 20,000

The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. (My rock & roll bruised eardrums can't hear much past 10K)

Cats 100- 32,000
Dogs 40- 46,000
Horses 31- 40,000
Elephants 16- 12,000
Cattle 16- 40,000
Bats 1,000- 150,000
Grasshoppers and locusts 100- 50,000
Rodents 1,000- 100,000
Whales and dolphins 70- 150,000
Seals and sea lions 200- 55,000

You guys really should know better. The point of being able to record frequencies far above the limit of human hearing is to preserve the harmonics. The lack of higher harmonics affects the perception of what you can hear.

The reason a decent record player kills digital is that a good setup will reveal frequencies into the 100,000 Hz range. The sound is much more natural.

Apart from that i have one word that should get your attention ... headroom.

[GlennChan]

The sound department might decide to do seperate sound for various reasons, especially since you can sync timecode with RED.
Hmm you might have to be careful about the workflow here?? AFAIK, Redcine (when it is released) does not have a feature to sync audio to timecode. The camera can record external timecode, but Redcine can't take external audio and sync it to your video. Nor can it slip the audio into sync if the timecode has drifted.

On crossing the line, I believe the workflow they used is the low-tech method... use a slate, find the frame where the clap hits, find the spike in the audio waveform, and manually line up the audio. This process takes the most time.



Going back a bit, as soon as the footage was loaded in we copied the drive of footage onto another empty drive so that we had two copies and my assistant editor could sync the production sound to the rushes on another machine (a macbook pro). We were under a great deal of time pressure so we started cutting before the sound was synced to the picture, and we just grabbed sound as we needed it. Then when the cut was locked we went through and added in the sync production sound.
http://www.reduser.net/forum/showthread.php?p=31942#post31942

Also see...
http://www.reduser.net/forum/showthread.php?t=3113

2- Perhaps someone else could chime in here.

3- The easy workflow would be to record audio onto the camera.

[Oskari Sipola]

Personally I'd find that for features or other high-end production purposes a separate recorder would be best. If for no other reason then because the cameraman doesn't want the soundguy to be tethered to the camera. Shoot it like you would shoot film; both sound and picture benefit. This is how feature productions are done, even if they are done with HDCAM or Digibeta, which both can include sound.

The sound features of RED come to use when doing ENG-style productions, and with four different channels it is a vast improvement to most tape-based ENG solutions (namely DVCAM and DVCPRO). Being able to record an on-board mic, a separate shotgun and two wireless lavaliers adds versatility to post, and is going to be much obliged by the sound mixers at your company. No more downmixing!

[Mark L. Pederson]

AFAIK, Redcine (when it is released) does not have a feature to sync audio to timecode. The camera can record external timecode, but Redcine can't take external audio and sync it to your video. Nor can it slip the audio into sync if the timecode has drifted.



Glenn, do you know for a fact that "Redcine (when it is released) does not have a feature to sync audio to timecode" ??

I am ASSUMING it will be the case ... and I will continue to beg for this feature until someone says "shut up Mark, we're not going to do it" ...

BWF sync support in REDCINE please.

Thanks in advance!

[Martin Drew]

The reason a decent record player kills digital is that a good setup will reveal frequencies into the 100,000 Hz range. The sound is much more natural.


I disagree. I believe the reason a decent turntable can be more enjoyable to listen to is because of the pleasant distortion created, not the frequency response. Most modern recordings will go through at least one digital stage before they end up on vinyl, but some people still prefer listening to playback from vinyl rather than digital. Ultimately it is subjective.

M

[PenGun]

I disagree. I believe the reason a decent turntable can be more enjoyable to listen to is because of the pleasant distortion created, not the frequency response. Most modern recordings will go through at least one digital stage before they end up on vinyl, but some people still prefer listening to playback from vinyl rather than digital. Ultimately it is subjective.

M

Well the $10,000 worth of stereo sitting on my floor says different. The digital wall, cutoff at 20,000 Hz produces the nastiest distortion I hear. A good record with a decent ginsau knife stylus in a good MC cartridge reveals way more information than any digital recording.

Your consumer POS record player is not in this class BTW.

A good phono section is never cheap and always better feeding rather than being in a decent preamp which again cannot be made cheaply. Damn a decent capacitor cost a lot to make.

Oh well I'm an audio freak so just ignore me.

[GlennChan]

Glenn, do you know for a fact that "Redcine (when it is released) does not have a feature to sync audio to timecode" ??

I am ASSUMING it will be the case ... and I will continue to beg for this feature until someone says "shut up Mark, we're not going to do it" ...
Hey Mark, I'm assuming just like you. If it did let you sync audio... then presumably someone from the Red team would say that the feature will be in Redcine.

2- Why stop asking for the feature if someone tells you it's not going to happen? ;)

[Martin Drew]

Well the $10,000 worth of stereo sitting on my floor says different.

In what way is the price of your stereo relevant?


The digital wall, cutoff at 20,000 Hz produces the nastiest distortion I hear. A good record with a decent ginsau knife stylus in a good MC cartridge reveals way more information than any digital recording.

More information than the digital source that was involved in the recording? Or do you chose only to listen to recordings which have no digital element in the recording?


Your consumer POS record player is not in this class BTW.

Do you know me? How do you know what equipment I have?


A good phono section is never cheap and always better feeding rather than being in a decent preamp which again cannot be made cheaply. Damn a decent capacitor cost a lot to make.

True. A good quality phono preamp is much much more expensive than a line level preamp. One reason why vinyl playback is much more expensive to impliment well than digital playback.


Oh well I'm an audio freak so just ignore me.

The implication being that I just won't understand, or what?

The style of your post seems quite arrogant to me, I made a fair point which you haven't addressed. I disagreed with your catagoric statement, I explained why and provided a logical example to illustrate my point. I love audio but I hate irrational and unthinking argument. If you measure the quality of your audio by how much your hifi costs then might I suggest that you try using solid gold interconnects, then it must sound better... right?

M

[PenGun]

In what way is the price of your stereo relevant?

It's my reference.



More information than the digital source that was involved in the recording? Or do you chose only to listen to recordings which have no digital element in the recording?


But of course. Although there are a few digital masters I can enjoy.



Do you know me? How do you know what equipment I have?


I am sorry. I did not mean your personal player. I was using the "your" colloquially. I was not clear.



True. A good quality phono preamp is much much more expensive than a line level preamp. Onr reason why vinyl playback is much more expensive to impliment well than digital playback.


Yup. It's a big gain and it's very hard to get right.



The implication being that I just won't understand, or what?


Yes really. Not many people have heard a good record player set up well with a great record on it.



The style of your post seems quite arrogant to me, I made a fair point which you haven't addressed. I disagreed with your catagoric statement, I explained why and provided a logical example to illustrate my point. I love audio but I hate irrational and unthinking argument. If you measure the quality of your audio by how much your hifi costs then might I suggest that you try using solid gold interconnects, then it must sound better... right?


No actually gold is only useful because it does not tarnish. It's 66% as conductive as silver which is 100%. Copper is in the high 90% range.

I mostly have rhodium plated connectors.

You can disagree all you want but the fact that a good turntable and system can deliver 100,000 Hz to my tweeters which make a very big difference in the reality of the sound. The ear is very sensitive to harmonics and they all ring against each other. When the upper harmonics are missing the lower ones just don't sound right.

There is nothing irrational about high end sound. It's one of our most acute senses and is not easily fooled. Most recorded music is very different than live performance and some of us try to narrow that gap a bit.

Really. Ignore me, I'm crazy.

[Jeff Kilgroe]

PenGun,

Take it down a notch... Most of your posts to these forums have been argumentative or instigating.

I moved this into the OT section because that's exactly what this discussion has become.

[Michele Gavazzeni]

This is supposed to be audio recording for cinema or video.
dialogs not a martin guitar or a Steinway&Sons piano!
Harmonics in a dialog? captured using shotgun mics?

[Martin Drew]

Most recorded music is very different than live performance and some of us try to narrow that gap a bit.


At least we can agree on that. You chose the analogue route I chose digital, as I said it is all subjective.

I won't say anything more on the subject because we have moved too far off topic.

M

[PenGun]

At least we can agree on that. You chose the analogue route I chose digital, as I said it is all subjective.

I won't say anything more on the subject because we have moved too far off topic.

M

Peace brother.

[Mark L. Pederson]

Just for you guys go -

http://www.digitalartsonline.co.uk/news/index.cfm?NewsID=8504

[Jack Wester]

Just for you guys go -

http://www.digitalartsonline.co.uk/news/index.cfm?NewsID=8504
How are you going to play a .mov file without going digitial?

Beats me :clown2:

[PenGun]

The only reason I brought up the record player thing was to illustrate the need for headroom. 48KHz is not enough. Digital needs all the headroom it can get, We write a new word every time we edit and 48,000 slices/s aint enough.

The test is crazy you want us to listen to a digital file and tell if it was produced with a digital or analog chain. Where did that file originate? Looks to me like they are all digital files originally anyway. The brick wall at 20,000 Hz is still there. There are artifacts dithering to that wall and no overtones. I think it would be very hard to tell anything actually. Probably why the test was set up.

A better test I have. I have Zappa's 'Man From Utopia' in both the original, all analog and the digitally remastered record. There is a definite difference and although the remaster is as good as I've heard from that era it does not come up to the old analog version to my ears.

[kmikami]

The only reason I brought up the record player thing was to illustrate the need for headroom. 48KHz is not enough. Digital needs all the headroom it can get, We write a new word every time we edit and 48,000 slices/s aint enough.

You have proven that you don't know what you're talking about. Sample rate only effects the upper frequency that can be recorded. It has no effect on headroom. Bit depth on the other hand effects the signal to noise ratio and therefore determines the amount of headroom you have. And 24 bit certainly leaves plenty of headroom for any application you can come up with, considering that the noise floor of a good 24 bit converter is lower than most if not all mics and preamps.

[GlennChan]

How are you going to play a .mov file without going digitial?
Jack... they are testing a subset of "analog versus digital"... that is, they are testing mixing between an analog SSL console and mixing in Pro Tools. They aren't testing analog versus digital recording. And it is somewhat implicit that the final format will be a digital format (e.g. you master a CD out of that recording)... because the majority of music released now is on digital formats.

In the console, they are also testing the signal processing... e.g. EQ and compression.

[Jack Wester]

Jack... they are testing a subset of "analog versus digital"... that is, they are testing mixing between an analog SSL console and mixing in Pro Tools.
Yes, but if PenGun was right in regards of the capping issue, it would not matter where in the process the capping took place. Once its gone, its gone making the experiment unrelevent as an argument to prove PenGun wrong.

That is not to say his right. It simply means that the test does not prove him wrong.

[Jack Wester]

Don't consider this as a thread hijack, its just a small parenthesis and this thread lost track a long time ago anyway ;-)

I've placed an order on a Sound Devices 702T together with a Shoeps shotgun (CMIT5U) and the Shoeps MK41 + CMC6 + CUT1.

Now I want to get rid of my recently purchased Rode NTG-2, my Rode aluminum boom pole and my Zoom H4 (2xXLR, 96KHz, 24bit recorder). I purchased the stuff two weeks back and I've only played with it in my home. Anybody intrested? If you pick them up for cash in Stockholm you can have them really cheap.

And heres "no highjack" part two:

I want to get a light weight carbon fiber boom pole. Anyone suggestions? Ambient? K-Tek?

[Gavin Greenwalt]

Correct me if I'm wrong but it seems like you wouldn't necessarily be able to get 24khz audio out of 48khz sampling. What if your sample rate is perfectly in phase with a 24khz signal? Theoretically you could sample at the exact same amplitude every single cycle resulting in a flat soundwave. And unless you somehow recorded perfectly in phase you would never accurately measure the magnitude properly.

It's something I've never understood about audio sampling discussions. What am I missing?

[kmikami]

Correct me if I'm wrong but it seems like you wouldn't necessarily be able to get 24khz audio out of 48khz sampling. What if your sample rate is perfectly in phase with a 24khz signal? Theoretically you could sample at the exact same amplitude every single cycle resulting in a flat soundwave.

To accurately reproduce a frequency you need to sample it twice which is why the sample rate needs to be twice the highest frequency you want to capture. So if you're sampling a 24khz signal you're not sampling the exact same amplitude every cycle. You're sampling exactly the peak and the trough and can therefore reproduce a perfect 24khz sine wave.

In practice there can be some messiness in the filtering that will effect the higher frequencies. So you can either buy expensive converters with well designed filters or you can buy cheaper 96k converters which push the filtering issues up into the inaudible range and waste a bunch of bandwidth and storage space. Your choice.

[Kevin Halverson]

Correct me if I'm wrong but it seems like you wouldn't necessarily be able to get 24khz audio out of 48khz sampling. What if your sample rate is perfectly in phase with a 24khz signal? Theoretically you could sample at the exact same amplitude every single cycle resulting in a flat soundwave. And unless you somehow recorded perfectly in phase you would never accurately measure the magnitude properly.

It's something I've never understood about audio sampling discussions. What am I missing?

Nyquist dictates that the amplitude must be zero at Fs/2 so 24 kHz audio out of a 48kS/s system is a clear violation and impossible. In general, with high order anti aliasing filters about .42 Fs is the practical upper limit of most quantizers used in digital audio applications. This is the trade off between alias rejection and bandwidth. Good engineering design practices (audiophile listening testing too) dictates that aliases must be avoided at all cost.

My "day job" was in the highend hifi industry and I spent over 20 years designing both recording and playback hardware for consumer and professional applications. Despite the rather good information available, there seems to still be a disconnect when it comes to sampled systems, so don't feel bad about your lack of "understanding", your in good company.

[kmikami]

Ah, ok. You guys are right. Nyquist/Shannon says that the sampling rate must be greater than the highest frequency to be sampled. So strictly speaking 24kHz is impossible but 23kHz is fine.

[Kevin Halverson]

23 kHz really isn't fine at all. In order to achieve adequate aliases rejection at 24 kHz and passband performance at 23 kHz, you would require an amazingly steep filter. To achieve adequate 16 bit performance you would need a filter that had 16 x 6.02 dB attenuation by 24 kHz or -96.32 dB, for 24 bit performance it would be over -148 dB. This isn't practical at all in the span of 1 kHz change, in fact it is nearly impossible. To achieve adequate aliases rejection, the highest practical passband performance is about .42 Fs or in the case of a 48kS/s system, its near 20.16 kHz for a 24 bit system. For a 16 bit system, it can be relaxed a bit, but even then an achievable upper boundary is about 21.6 kHz.

[GlennChan]

Nyquist dictates that the amplitude must be zero at Fs/2 so 24 kHz audio out of a 48kS/s system is a clear violation and impossible.
I don't think it's impossible (depending on what context you're coming from). You can reproduce a 24khz signal... you're just not sure if it's the right one. Different 24khz signals will yield the same sampled values... working backwards from there is ambiguous. In a theoretically perfect system, you don't want aliasing and you would filter off 24khz frequencies before they hit the sampling device. So I suppose it's "impossible" in that sense... but it's not strictly impossible no?

2- In practical systems everything is different since you need to consider aliasing, (possibly ringing artifacts?), and keeping the design reasonable.

3- And on a totally different note... if you wanted to get good sound with Red, I think a good place to start would be:
Boom mics- a hypercardioid and a shotgun
All the accessories for that (boom pole, windsock, etc. etc.)
2 wireless mics (with lavs, and mics you can plant/hide)
3/4-channel mixer (*though I don't know the best way of monitoring, since you'd ideally want to check the return on all channels; and you want to output 4 channels, but the SD stuff only outputs 2 channels AFAIK)

This is preferable to no wireless mics and recording double system sound. Because with wireless, you can get your mic a lot closer and this will improve sound quality dramatically.

[Kevin Halverson]

Yes, it really is impossible. With a 24 kHz signal sampled by a 48 kS/s system you will have two samples with the exact same amplitude. This is the same as DC, their will be no resultant signal, period. Now if the applied signal is anything (and I mean anything at all) above 24 kHz it will fold back to become an alias. For example, if you apply 24,001 Hz to a 48 kHz sample system, what will result from the sampled system is a 1 Hz signal with an amplitude identical to the 24,001 Hz one that is applied.

In sampled systems, the Nyquist limit is an absolute. Not really close, but an absolute. This may not be all that easy to wrap your head around, but rest assured, my comment is absolutely correct.

[GlennChan]

A 24Khz signal would alternate between two values no? (And it's possible that it would alternate between 0 and 0, but that's not what I'm talking about.) At those two values, if they are not the same, will create a signal. And in rare cases it will be the original signal that is reproduced.

[Kevin Halverson]

A 24Khz signal would alternate between two values no? (And it's possible that it would alternate between 0 and 0, but that's not what I'm talking about.) At those two values, if they are not the same, will create a signal. And in rare cases it will be the original signal that is reproduced.


NO!

It will alternate between two samples, each with identical amplitude. The magnitude of the two values can be anything between + to - full scale, but they will always have identical amplitude and will represent zero signal.

Consider the following. The period of a 24 kHz waveform is 1/24,000 or 41.6 us. The two samples will be at 1/48,000 or 20.83 us. Regardless of the starting phase relationship between the applied signal and the sampled system, the two samples will be offset by 180 degrees and will have the exact same value.

[PenGun]

Ah yes the joys of digital sound. So we write a word 24 bits long 48,000 times a second. Slicing and dicing this profoundly analog medium in an attempt to make something sound good.

Then we take that pile of 24 bit long words and edit them, losing a bit or two every time. With luck you could get to something approaching CD sound.

You really are satisfied with CD level sound for this amazing camera? I'd do 192K myself and probably still not be very pleased.

You gotta do at least 24/96.

[Kevin Halverson]

Having been involved in the manufacture of very high end analog audio components for nearly three decades now, I can appreciate your concerns. However, I can state with absolute confidence, that if properly quantized, processed and reconstructed, the process can be very benign in terms of the harm that it imparts.

The advantage that higher sample rates provides isn't the added bandwidth (which only has a minor contribution), rather it is the relaxation of the requirements of the anti alias and anti image filtration that a properly executed sampled system dictates.

I have designed a number of A>D systems that operate well above 192 kS/s and with proper care, this signal can be stored at much lower sample rates (like 48 kS/s as an example) and still remain so pristine that is nearly impossible to differentiate from the original. I sit on a number of the TCs for advanced audio formats and have been involved in their implementation long before they were ever released as consumer and professional formats.

What I am attempting to convey is that don't assume that a really well executed system that stores LCPM at 48 kS/s with 24 bit samples can't be extremely high fidelity and very faithful to the original. Its really how its accomplished. As for the RED One, who knows how it will perform, but I can assure you, I will certainly be running it through a battery of tests (both objective and subjective) as soon as one is available to me.

[Martin Drew]

Ah yes the joys of digital sound. So we write a word 24 bits long 48,000 times a second. Slicing and dicing this profoundly analog medium in an attempt to make something sound good.

Then we take that pile of 24 bit long words and edit them, losing a bit or two every time. With luck you could get to something approaching CD sound.

No. Thats the thing with digital, once it is in the digital realm it is easy to retain quality. When you edit 24 bit files they remain 24 bit, you will not lose any bits by editing. Quality is determined most significantly at aquisition in the digitisation stage, once it is digital it is all just numbers.

I would be most concerned about the engineering of the input stage and digitisation, I think 48 Khz is acceptable if this is implimented well. It would be nice to have 96 Khz but higher sampling rates require compromises, going higher than 96 Khz may well reduce quality mot improve it.

M