Forums | Search | Archives

All Forums  Hardware & Soundcards  True 24-bit recording with Terratec EWX and DMX Fi

Author	Topic
outerspacelord Posts: 31	Posted - Mon Mar 03, 2003 6:58 pm  Here is some info that may be of interest. I have a Terratec EWX 24/96 tested with WDM Driver v5.0.2000 Build 126 in both Win98 SE and XP SP1. This card use the same drivers as Terratec DMX Fire 24/96, so read on ... First in XP I set driver in Cool Edit to 4-byte IEEE float and did a stereo 32-bit (float) 96kHz recording (with nothing connected to the input). Then I ran the Analyze\Statistics function in Cool Edit (before saving the file), and the interesting thing i now see is that actuall bit depth is 16 bit. That is bad ... Ok so then i try the same again changing 'Get 32-bit audio using: 4-byte PCM' in device properties of Cool Edit. Now the actuall bit depth says float in Analyze\Statistics. Very good ... or ? It shows that when I connect a signal to the input and record it, nothing get recorded. So this does not work at all in XP. Now I set get 32-bit to 3-byte Packed PCM in Cool Edit and record again. Now the actuall bit depth says 24 bit. So this is the working mode and hopefully the right setting ;) So now to the difference between XP and Win98 SE. First 4-byte IEEE float record some load and strange noize (with no signal on input). Means noneworking ... Then 4-byte PCM behaves just like 3-byte Packed PCM in Win98 reporting actuall bit depth to 24 bit. So it seems like 4-byte PCM and 3-byte Packed PCM both are working the same way in Win98? If I now choose 3-byte Packed PCM as playback too it says in lower left in Cool Edit that it is playing as 24-bit. So here I choose 4-byte IEEE float or 4-byte PCM on playback settings on device to get rid of this message. Here is some questions i would like to have answered. Audio Options in SONAR shows 4 bytes left-justified in Driver Profiles. What does this mean compared to all other modes discussed earlier? In RMAA 4.3 when I Test supported modes (Stereo) all shows D - full duplex mode (without 192 kHz). Is this good or bad? What does it mean? The test in Cool Edit shows that record should be set to 3-byte Packed PCM to actually work and to be able to record 24 bit. Both EWX 24/96 and Terratec DMX Fire 24/96 says: Bus master transfer supports '24bit 4byte mode' (32bit) at http://www.terratec.com/products/ewx2496/ewx2496_details.htm Is this the 4-byte PCM in Cool Edit and RMAA 4.3? Because if it is, it is not working at all recording in XP. I can't find this very consistent and is hard to get. 3-bytes this, 32 bit PCM and 4-bytes that
1wingman Posts: 47	Posted - Tue Mar 04, 2003 8:28 am  I'm gonna be really upset if 4-byte IEEE float isn't really 32 bit. I've spent hours working on a record thinking it was 32-bit float. (I just installed the EWX 24/96. I'm not sure what version the driver is, but I downloaded it from their site.)>:K>:K
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 11:30 am  Beaware that the '24-bit Alt' option in WaveLab also communicate with the driver in a way that the files actuall bit will be only 16-bit. Choose the '24-bit' instead for terratec's drivers.
SteveG Location: United Kingdom Posts: 6695	Posted - Tue Mar 04, 2003 12:24 pm  Be interesting if somebody else can replicate any of this - I haven't got either of these cards here - I have to rely on my Nephew, who has a 6fire, for information. But WDM on Win98SE is known to be problematic, and not recommended. But it is a bit strange about 4-byte IEEE float, because this is the floating format that CEP uses. Why it wouldn't import this format directly and correctly, I don't know. _________________
1wingman Posts: 47	Posted - Tue Mar 04, 2003 1:33 pm  I just got home and checked it out. Under CE's statistics for actual bit-depth it simply said "float". I don't know if that means it is 32-bit like CE's status bar says, or if it is just 16. Note, this is a file that has already been saved as 32-bit float. I've got to leave again, but I'll play around and try to duplicate what outerspacelord is seeing.
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 2:06 pm  1wingman: Like I said I did the analyze before saving the file. But you can also save and try import to wavelab and run 'global analysis' and look under 'Extra' and check what 'Apparent Bit Resolution:' says. I guess if you apply some edits on it may go float so don't use such a file. SteveG: I liked what you wrote about MAC's ;) I now have written e-mail to terratec support and maybe we'll see and answere from them in six months or so ... hehe I also read in the manual for DMX Fire 24/96 where it claims 32-bit communication with the driver. But it did not really make me any wiser. What I will do is to take out every pci card so to make sure my EWX 24/96 not sharing any irq's and let you know if all comes down to irq sharing.
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 4:43 pm  The IRQ sharing is now ruled out to not be the problem. Got my EWX 24/96 on IRQ 5 without sharing. All the same as written above. Also tried not using 'try as WDM'. What happens is that now it also reports 16-bit only in 3-byte Packed PCM too. So no help there. So for now it seems that '3-byte Packed PCM' with 'Try as WDM' enabled is the only solution to record true 24-bit.
SteveG Location: United Kingdom Posts: 6695	Posted - Tue Mar 04, 2003 5:15 pm  Quite frankly, the best way to determine the file type you've got is to enable 'show file information' when you open the file. This actually reads the file header info and will tell you everything about it, including the bit depth and coding type. And this is purely OS-dependent, not reliant on CEP, Sonar, or any other app. _________________
1wingman Posts: 47	Posted - Tue Mar 04, 2003 5:25 pm  I just recorded a short piece and used CE's analysis before saving. It reports "float" under bit-depth. The file headers from my saved waves indicate 32-bit float format also. I'm using driver version 5.0.2000.126 dated 9/18/2002 that I downloaded from their site. (The supplied driver didn't work.) I've got XP HE SP1. It seems to be working correctly for me. The only other thing that I can think of is that you might not have the EWX driver selected for use in edit view (EV). EV should appear beside the driver you have selected under device settings.
SteveG Location: United Kingdom Posts: 6695	Posted - Tue Mar 04, 2003 6:03 pm  If you have 'float' reported, it is a 32-bit file. But since CEP still supports multiple 32-bit file types, there is no guarantee which type it will be, and the only sure-fire way to tell is to do what I suggested above, or go View>Wave Properties>File Info. _________________
SteveG Location: United Kingdom Posts: 6695	Posted - Tue Mar 04, 2003 6:08 pm  Quote: I guess if you apply some edits on it may go float No it won't. The only way you can change a file type is to do it deliberately, unless it's an MP3, when it will be converted on the fly to a PCM type. _________________
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 8:11 pm  1wingman: What I believe you have done is to enable 'Adjust to zero-DC when recording'. Try not to do that. If it still reports float in analyze you are lucky. This will mean that there is a differece between EWX and DMX. Here is what I know about that: www.tomshardware.com/video/20020115/terratec-01.html Enabeling 'Adjust to zero-DC when recording' is the same as applying an effect and are done in software. This will not give you a true result of what i'm talking about. Think I read somewhere here why to turn off stuff like that while recording: www.rme-audio.de/english/index.htm By the way I have EWX driver selected for use in edit view (EV) Disabled all others, taken out all other cards, disabled tweaks and so on, hehe. SteveG: About View>Wave Properties>File Info: Here is what the helpfile for the 'global analysis>Extra' says "Apparent Bit Resolution displays the apparent bit resolution in the analysed selection. For example, you may have received a 16 bit file from somewhere, but it is very noisy. You suspect that the file was actually recorded with only 8 bit resolution and then expanded up to 16 bits. Run it through this analysis and you will know". This is exactly what the 'actuall bit depth' in Cool Edit also does. And this is more true than what file info does which of course report the format the file is saved in. What you can do to see for yourself: Digitally extract a song from an audio cd 16/44. Open it as 32-bit (settings>data) and save as '32-bit Normalized Float (type 3) - Default'. Now check the file with both wavelab and cooledits analyze functions and you'll see 16-bit. Ok, now do a plain 'Effects>Amplitude>Amplify...>Center Wave' and save again. Cool Edit will report float and WaveLab will report 36-bit. File info of course in both cases report 32-bit. My next test will be a older driver, and if not working I will test on another computer and also try a complete reinstall of the os. I always have 'Settings>Data>Interpret 32-bit PCM .wav files as 16.8 float (compatibility with old Cool Edit format)' disabled by default because of .pk files otherwise not saved on 32-bit files. Yes, it did not help to reenable ;) I also have DirectX 9 installed on this machine but don't believe this is the reason for the trouble. But we never know.
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 9:11 pm  Tested WDM Driver v5.0.2000 Build 124 and back to WDM Driver v5.0.2000 Build 116. Cool Edit behaves the same. So it was no help in that. Should soon start get sallary from terratec for all this hours of testing
1wingman Posts: 47	Posted - Tue Mar 04, 2003 9:35 pm  Thanks, now I'm gonna cry. It says 16 bit.>:K>:K
outerspacelord Posts: 31	Posted - Tue Mar 04, 2003 10:11 pm  1wingman: Please don't cry Set driver to 3-byte Packed PCM ... and 24-bit you go. But yes I know it won't help for your old recordings. I gently weep too. Set the 'Adjust to zero-DC when recording' as you want. Because now you knows what really happens. zero-DC will use some extra cpu and will be only trouble on slow computers. But it will not let you see the real stuff
SteveG Location: United Kingdom Posts: 6695	Posted - Wed Mar 05, 2003 4:33 am  outerspacelord wrote: SteveG: About View>Wave Properties>File Info: Here is what the helpfile for the 'global analysis>Extra' says "Apparent Bit Resolution displays the apparent bit resolution in the analysed selection. For example, you may have received a 16 bit file from somewhere, but it is very noisy. You suspect that the file was actually recorded with only 8 bit resolution and then expanded up to 16 bits. Run it through this analysis and you will know". This is exactly what the 'actuall bit depth' in Cool Edit also does. And this is more true than what file info does which of course report the format the file is saved in. What you can do to see for yourself: Digitally extract a song from an audio cd 16/44. Open it as 32-bit (settings>data) and save as '32-bit Normalized Float (type 3) - Default'. Now check the file with both wavelab and cooledits analyze functions and you'll see 16-bit. Ok, now do a plain 'Effects>Amplitude>Amplify...>Center Wave' and save again. Cool Edit will report float and WaveLab will report 36-bit. File info of course in both cases report 32-bit. Yes, but that's the whole point, isn't it? You can't determine what the file type actually is unless you interrogate the header, and you can't do this from 'analyse'... _________________
1wingman Posts: 47	Posted - Wed Mar 05, 2003 8:03 am  outerspacelord wrote: 1wingman: Please don't cry Set driver to 3-byte Packed PCM ... and 24-bit you go. But yes I know it won't help for your old recordings. I gently weep too. Set the 'Adjust to zero-DC when recording' as you want. Because now you knows what really happens. zero-DC will use some extra cpu and will be only trouble on slow computers. But it will not let you see the real stuff I didn't cry, but I'm very frustrated. Oh well, I'm glad you posted your findings here. Otherwise I would have been *happily* recording at 16-bits. I think I'm going to just finish burning the current record to CD, I'm at that point anyway. It took me forever to go through and remove the periodic clicks and pops. I will go back and record a sample at 24-bits to see if it makes enough of a difference to make me want to redo everything. I hope using the 3-byte packed PCM doesn't cause any performance issues. I guess it shouldn't on fast machines. Also, let me know what Terratec says about the issue. I'm going to ask them about it as well. Thanks
outerspacelord Posts: 31	Posted - Wed Mar 05, 2003 10:43 am  SteveG: I am not sure if I get you? The whole point is actually determine if the communication between Cool Edit (and other apps) and the hardware utilises the chip in the soundcard the right way so that it uses the claimed ability to record 24-bit recording on a analog input. And not only 16-bit. As you know you can record 32-bit files with a 16-bit soundcard too. The reason why to work in a higher domain like say 32-bit is explained very well in this essential reading http://www.terratec.de/4G/2496-en.pdf (which again is another topic in it self and which is discussed over and over again in this forum with very diffrent luck since it a lot of strange ideas around it . Everyone should read that document on their bed over and over again! David Johnston is the man over all, but where is he now when we seem to need him ?
SteveG Location: United Kingdom Posts: 6695	Posted - Wed Mar 05, 2003 3:18 pm  24-bit soundcard encoders always work at 24-bit, as far as I'm aware - you can't stop them. All that changes is the format of the output, which is determined within, and processed by, the driver, according to instructions passed to it from the software that is utilising the stream from it. Yes, you can record a 16-bit stream as 32-bit, because all that happens is that two extra empty bytes get added to each dataframe. All that happens if you record at 16-bit resolution with a 24-bit card is that the 24-bit output stream from the A-D is truncated at 16-bits - the other 2-3 bits of potentially useful data, and all the rubbish in the 3 LSB, are thrown away. And your 'essential reading' is a masterpiece of bloodcurdlingly awful translation from German. If this guy, whoever he is, actually knows what he's talking about, it doesn't exactly show. The bit that really cracked me up was his explanation of the floating point technique, which, even allowing for a piss-poor translation, leaves out the essential bits about how scaling works completely! And there's that spectacular leap into quatisation in the middle of it, which is, to say the least, misplaced... Yes, Terratec are a German company, and their ability to communicate in English, and their apparent desire not to, are legendary. If you can get any of their agents to tell you the truth about this, it's quite revealing... But I couldn't honestly recommend this document as a model of clarity at all. There are some far better english documents around. _________________
1wingman Posts: 47	Posted - Wed Mar 05, 2003 7:41 pm  Their manual was fun to read, almost "freaky"!:D Steve, are you saying we might not get an answer to this problem?
1wingman Posts: 47	Posted - Wed Mar 05, 2003 10:14 pm  More info. I found that when using 4-byte float @ 96khz (with 96k selected in the EWX control panel) the sample rate in the EWX control panel changes from 96k to 41k once the record button is pressed in CE. I don't know why or what this means, but it does appear that the recorded wave was sampled at 96khz based on resulting file size. The control panel has a button labeled "Samplerate Locked" in the Master Clock group. Toggling this button to "on" and recording in CE with the device wave in set to 4-byte float results in a file with no signal, but if you analyze it the bit depth is reported to be 24. The sample rate also stays at 96k in the control panel. This is pretty much what you get if you try using 4-byte PCM as the wavein setting. I also discovered that you can record using 4-byte PCM if "Try as WDM" is unchecked. This results in a file with a bit depth of 16 though. There is also a brief buzzing sound when you press the record button. If I remember correctly, it doesn't show up in the wave file itself. The only thing that appears to work correctly is recording with 3-byte packed PCM. The interesting thing about this option though is that if you save the wave immediately after recording without applying any functions to it, when you reopen it CE reports the bit depth to be 32 in the open file dialog. Analysis reveals 24 though. Once you apply a function like Click/Pop removal, save and reopen the file, the bit depth is "float" under analysis.
SteveG Location: United Kingdom Posts: 6695	Posted - Thu Mar 06, 2003 2:46 am  Quote: CE reports the bit depth to be 32 in the open file dialog. Analysis reveals 24 though. Of course it will on an untouched file! The A-D is 24-bit! You will get 'float' displayed when the bit depth is greater than this, because the actual resolution gets quite tricky to determine at greater depths. It is a relatively simple operation to determine the actual bit depth of an integer file, but strictly speaking, you will only get 24-bit resolution from a floating-point file, although the variable scaling factor means that you can have this 24-bit resolution ocurring at different reproduced levels. That's what the German blue-eyed boy failed to explain. Come to that, reading between the lines, I don't think he actually understands how a digital signal is reproduced in the analog domain, judging by what he's said about it. We had a long discussion about this here a while back - there are some very common misconceptions about this, and he seems to have picked up at least one of them (as far as anybody could ever actually tell!) It strikes me, though, that all of these difficulties seem to relate back not to the soundcard, but the driver, and its interaction with the OS and software running under it. It's very hard to tell what's really going on, because there are actually too many potential variables at work. Really, the only hope of clarification rests with Terratec. But I wouldn't hold your breath. Ask the TerraTec agent in your area to see if he can find out. He has to email Germany. They have to translate his request, interpret it and then translate it back again (if they respond at all). And you've seen what these translations look like. Do you really think we're ever going to find out? _________________
outerspacelord Posts: 31	Posted - Thu Mar 06, 2003 2:50 am  SteveG: Yes, I know the germans did not win the last war, but ... And I know the germans seems not willing to start war this time when their asked to. My native language is not english so I may be a little more merciful than you. Hehe. Ok I agree that dude look a bit funky so to say ... bad hairday? Anyway, I believe that the document is helpfull for some kids like me because there is so much superstition around high resolution audio like 24/96. Just take a look around in this forum? At least by reading that document you will be convinsed it is a good thing even if you don't understand a thing, all without being a professor in math. Can you please recomend some of your 'essential reading' stuff please?
outerspacelord Posts: 31	Posted - Thu Mar 06, 2003 3:25 am  A very important thing I forgot to mention is that I am very happy with my card. Low latency with softsynths and all that. And it is very easy to manage. So I have no problems reccomending it. I may even buy a DMX 6fire 24/96 cause it has midi and that breakout box. All this has been for trying to solve how to communicate the right way with the driver (as Stevie said) so to utilize the full potensial of the A->D. Hopefully we found out some? Or at least that we should stay alert. A week ago I bought a SoundBlaster Audigy 2 eX from Creative Labs ( the liers yes I know). Gave it a two hours try and gave it to my kids Seemed ok, but to much hazzle and I don't care much about DVD on PC and games and so on and on. But that's another topic. No, I never expected to get answere from terratec support ... They maybe busy making some new revolusionary 'german über alles' muzak. And that is kewl enough for me
outerspacelord Posts: 31	Posted - Thu Mar 06, 2003 3:47 am  quote: Of course it will on an untouched file! The A-D is 24-bit! Hey hey hey, that was my solution all the time from the first post ... We have to use 3-byte Packed PCM as record setting. If not no sound or only 16-bit it will be.
1wingman Posts: 47	Posted - Thu Mar 06, 2003 9:34 am  SteveG wrote: Quote: CE reports the bit depth to be 32 in the open file dialog. Analysis reveals 24 though. Of course it will on an untouched file! The A-D is 24-bit! You will get 'float' displayed when the bit depth is greater than this, because the actual resolution gets quite tricky to determine at greater depths. It is a relatively simple operation to determine the actual bit depth of an integer file, but strictly speaking, you will only get 24-bit resolution from a floating-point file, although the variable scaling factor means that you can have this 24-bit resolution ocurring at different reproduced levels. That's what the German blue-eyed boy failed to explain. Steve, I know that the A-D is 24 bit. It makes sense that analysis shows 24 bit before the file is saved as 32 bit float type 3. What I don't clearly understand is why the file analysis reports the bit depth as 24 after it was saved as 32-bit float type 3. I'm a bit rusty on floating point format so please forgive my next question. Why do you only get 24-bit resolution from a floating point format verses an integer format? Are you saying that there are disadvantages to having to get the audio from the card in 3-byte packed PCM format verses 4-byte float?
SteveG Location: United Kingdom Posts: 6695	Posted - Thu Mar 06, 2003 8:04 pm  1wingman wrote: I'm a bit rusty on floating point format so please forgive my next question. Why do you only get 24-bit resolution from a floating point format verses an integer format? Are you saying that there are disadvantages to having to get the audio from the card in 3-byte packed PCM format verses 4-byte float? No, not really. You are never going to get more than 24-bit significance from the soundcard, whatever you do, and you can send this info in 3 byte dataframes. You only need the 4th byte to store the 32-bit floating point result. But it doesn't contain individual 32-bit results, it only contains scaled 24-bit results - that's what 32-bit floating point data is. It comprises a 24-bit mantissa, a seven bit exponent, and a sign bit. The 24-bit mantissa always takes the same numerical form, but the scaling bits effectively determine the position of the decimal point in the answer that is output - this is how it can 'float'. But the amount of actual resolution in each word can still never exceed 24-bits. By some fancy mathematical footwork, it is possible to realise a dynamic range, using this method of ranging the output of the signal, of about 1500dB. And this is why you will find it virtually impossible (unless you take deliberate steps) to make a floating point signal clip! So the output values will be in the form of scientific notation, ie, a 24-bit number raised to a power - 1.28557395867211044823469 x 10^25, for instance. You want some better reading? Try this. It may not answer all the questions you might have (and you do have to be a little cautious of some of the stuff in the Bob Katz link, although his heart's in the right place), but the main body of what Haas is saying is irrefutable - and so it should be when you consider who's saying it! _________________
outerspacelord Posts: 31	Posted - Thu Mar 06, 2003 10:34 pm  Ok, thanks Steve ... Here is a few of my (math favorites: http://www.tonmeister.ca and from Center for Computer Research in Music and Acoustics http://www-ccrma.stanford.edu/~jos/mdft/Number_Systems_Digital_Audio.html http://www-ccrma.stanford.edu/~jos/pubs.html Quote: Are you saying that there are disadvantages to having to get the audio from the card in 3-byte packed PCM format verses 4-byte float? It seems to me that the only disadvantages is performace, which I guess is really not an issue since todays computers are very powerful. Probably it only put some more unmeasureable strain on the cpu. We should not really care, just be aware of it. As I showed earlier the '24 bit alt' alternative in WaveLab seemed to do exactly the same as 4-byte PCM in Cool Edit. It gave no input. So here is what WaveLabs helpfile says about it: "The "24-bit Alt" option is an alternate way of transmitting 24-bit audio, used by certain audio hardware. If you plan to work with 24-bit audio, you should consult your audio hardware's documentation to find out whether you should use "24-bit" or "24-bit Alt". If the documentation isn't clear about which method is supported (or if both methods are supported), you should try using the "24-bit Alt" mode, as this may give better performance." As you can see performance ... When you save the file in 32-bit the advantage for you still will be float giving you beter dynamics while processing. As the german dude said even som plugins work internally in 48-bit. Some of Sonic Timeworks plugins even work internally 64-bit if i'm not wrong ...
outerspacelord Posts: 31	Posted - Thu Mar 06, 2003 11:49 pm  off topic: Is it best to remove dc-offset from a 16-bit file before i convert it to 32-bit for editing, or after? Yes or No? And then why
SteveG Location: United Kingdom Posts: 6695	Posted - Fri Mar 07, 2003 1:55 am  outerspacelord wrote: off topic: Is it best to remove dc-offset from a 16-bit file before i convert it to 32-bit for editing, or after? Yes or No? And then why Depends how big the DC offset is. The really small ones of a fraction of a percent of baseline shift really don't matter very much, if at all - most waveforms have more asymmetry in them than this, anyway. If, for any reason, you have larger shifts, then they will have the effect of limiting either the positive or negative excursions of the recorded waveform. Strictly speaking, this isn't a problem with a floating point waveform, but it could potentially be with an integer one. The only thing you really have to bear in mind is that if you carry out any operations on your waveforms, especially ones that involve non-linear manipulation, like compression or expansion, that these will also have an effect on the DC shift, and could potentially (again!) cause miscalculations to occur in the processing. Linear effects are only going to cause baseline shifts in proportion to the original signal amplitude. So the bottom line really, I suppose, is to eliminate the DC shift before you convert, or do anything else at all. I'm not saying that this is magically going to cure all ills - it's really more like preventative medicine. _________________
outerspacelord Posts: 31	Posted - Fri Mar 07, 2003 2:13 am  I just love your yes and no answeres You will do good as politician ...
SteveG Location: United Kingdom Posts: 6695	Posted - Fri Mar 07, 2003 6:14 am  outerspacelord wrote: I just love your yes and no answeres You will do good as politician ... I don't think so. I'm not too good at declaring war on other nations. But you should always beware of simple answers to complicated questions, and vice versa. Despite what you might think, this isn't actually simple, but if you really want a yes/no answer, then you can have Yes. _________________
outerspacelord Posts: 31	Posted - Fri Mar 07, 2003 11:37 am  Quote: So the bottom line really, I suppose, is to eliminate the DC shift before you convert, or do anything else at all. I'm not saying that this is magically going to cure all ills - it's really more like preventative medicine. I read that as a small yes. hehe ... And I agree to your explanation. It is good people like you are around.
outerspacelord Posts: 31	Posted - Sun Mar 09, 2003 6:04 am  I now tried Egosys Waveterminal 2496 (http://www.esi-pro.com/) and found this card to report and work 'float' (with signal in) using '4-byte PCM'. Meaning it is working! '4-byte IEEE float' setting reports 16-bit and '3-byte Packed PCM' reports 24-bit, just as EWX 24/96. So people should be aware and check for their soundcards! I will try later with Audgy 2 eX.
1wingman Posts: 47	Posted - Wed Mar 12, 2003 8:49 am  How do you have access to all of these cards?
Aaron Edwards Posts: 14	Posted - Sun Mar 30, 2003 12:58 pm  It's interesting, but I did a recording using "3 byte packed PCM" and opened the recording in edit view, switched the vertical units displayed to "Sample Values", looked at the resulting values, and damn if they weren't all numbers between roughly +30,000 and -30,000, which is to say -32768 and +32768, or 16 bits! (The Analyze>Statistics window said the Actual bit Depth was "float", but only showed sample values between the above cited value range. What is up here? Is this a reporting error? Or does the software somehow "see" that I'm feeding it an analog output of a CD recorder and scale itself down to 16 bits, all the while cursing me for wasting it's time? I've been following this topic, but haven't seen any mention of looking at the actual sample values yet, so I'd like to know if anyone else had tried this, and, if so, what results they got.
	Topic

Page: