Opinion? Recording interviews with MP3 320kbps at 48kHz

[Inprogress]

What is the general consensus about recording in MP3 format nothing lower than 320kbps at 48khz? How does it compare to AC-3 specifically for voice recordings, i.e. use lav mic with "cheap" MP3 voice recorder for interview.

Just been wondering. I have a GH2 as well as a Tascam DR-40 for which I now have to upgrade microphones for (currently have a Rode Video Mic and Audio Technica ATR3350 Lav).

 

[Steve House]

MP3 is a distribution format, not a mastering format. Never use any lossy compression scheme for original master recordings. Best is 48kHz uncompressed WAV or BWF (broadcast wave - ie, a standard wave file with additional metadata in the file header), or if you really really NEED compression use a lossless compression format such as FLAC. The only reason to use a format such as MP3 is to save on storage space but storage is cheaper than dirt these days so it's a false economy.

 

[Inprogress]

Thank you Steve.

 

[Noiz2]

I largely agree with Steve.

Compressed formats can sound fine and one can make a pretty solid argument that there are currently no uncompressed formats, except maybe the 1 bit recorders. At the same data rate MP3 should sound better than WAV since it is a "smart" compressor that assignes more space to complicated sounds and WAV saves space by only looking at the signal every so often and just assuming what lies between points is a smooth curve and assignes the same amount of data to simple sounds as it does to complex ones. You can up the sample rate but your still locked into giving every sample the same amount of data.

But practically speaking nobody is using MP3 that way and so you would be recording at a data rate probably 1/10 or so of the WAV.

IF you have a good preamp as a front end and you push the level you can get good results, but as Steve mentioned with memory costing what it does there is no good reason to go there, except record time. IF you need to do extremely long takes as in you are recording a twelve hour ??? straight with no breaks then MP3 may be your only option since you may not be able to get a card big enough and it won't be cheap.

However you will need to convert that MP3 to WAV or AIFF before you use it.

And I would stay away from "cheap voice recorders" because they are very likely to have clocking stability issues and that will make syncing a living hell.

 

[Rick R]

+1 to what Steve and Noiz2 stated.

 

[zjrt02]

At the same data rate MP3 should sound better than WAV since it is a "smart" compressor that assignes more space to complicated sounds and WAV saves space by only looking at the signal every so often and just assuming what lies between points is a smooth curve and assignes the same amount of data to simple sounds as it does to complex ones.

Huh?

Unless I've been lied to for 20 years WAV files are sampled but NOT compressed. One may sample at a smaller word size and/or sample rate which will reduce file size (and fidelity) but the original captured fidelity remains within the confines of your chosen sample rate/size. An original 320 Kbps stream rate 16-bit 48kHz MP3, by it's very nature, will have less information than an original 16-bit 48Khz WAV of the same source. MP3 encoders, even at the highest stream rates, will always throw away data while applying "perceptual" encoding - trying to trick the ear into hearing something that's not there. MP3s are the JPEG of audio. JPEGs can look good but they're never as good as the RAW image captured from a camera sensor or original digital art work, and when you go to manipulate them they can cause big problems.

Fidelity, or how close the copy is to the original, in WAV files is a mainly a function of word size and sampling rate (ignoring the influence of the converters, etc.). No digital copy of an analog signal can be "perfect" but an original 24bit/96Khz WAV can come mighty close and you can mash it and bend it in post and it will still hold up. MP3s? Not so much.

Richard Thompson
T4 Media

 

[Richard Crowley]

Compressed formats can sound fine

Yes, they can, as long as that MPEG compression is the LAST step in distribution. If you intend to do any post-production editing, then compressed files can be quite HORRID and are to be avoided if at all possible.

and one can make a pretty solid argument that there are currently no uncompressed formats,

That is simply not true at all. At least by the conventional definition of "compressed" as we use it in the audio world.

except maybe the 1 bit recorders.

"1 bit" is only one of several analog-to-digital conversion (ADC) or digital-to-analog conversion (DAC) techniques. It is more properly referred to as "delta-sigma" technique. "1-bit" is largely a marketing term and has no significant superior benefits compared to other ADC or DAC methods. http://en.wikipedia.org/wiki/Delta-sigma_modulation

At the same data rate MP3 should sound better than WAV

I could not disagree with this statement more. EVERY lossy compressed bitstream (such as MP3) starts out as a "WAV file". Where I mean an original, unmolested stream of samples of the original waveform.

Lossy audio formats (such as MPEG, etc.) MIGHT sound virtually identical to the casual listener, but if you actually compare the signal (such as by inversion and cancellation) you will hear (and see if you are looking at an oscilloscope) the differences between lossy compression and the original signal.

since it is a "smart" compressor that assignes more space to complicated sounds

True.

and WAV saves space by only looking at the signal every so often and just assuming what lies between points is a smooth curve and assignes the same amount of data to simple sounds as it does to complex ones.

No. ALL digital audio (compressed and uncompressed) "looks at the signal every so often". Now "every so often" may be only 8000 times per second in the case of low-quality communications (speech) applications. Or it could be as high as 5644800 times per second (double-rate DSD). That is how we get continuous analog waveforms into digital form so we can record and edit the signals. Regular temporal sampling is the first stage of ANY and EVERY digital audio ADC and DAC method, technique, storage processing, transmission, etc.

You can up the sample rate but your still locked into giving every sample the same amount of data.

That is an accurate description of UN-compressed audio storage (eg. WAV, AIFF, files. etc.)
Then LOSSY COMPRESSION techiques such as MP3 take that "same amount of data" from each sample and throw away a large majority of that sample data. The details of exactly WHICH part of the data and HOW MUCH of it to throw away are the "magic" parts of the compression (and de-compression) algorithms of MP3 and all other lossy compression formats. Fraunhoffer has invested perhaps 100000 man-hours of research and development into this "magic" and has made perhaps billions of $£€ selling the implementation of those secrets.

However you will need to convert that MP3 to WAV or AIFF before you use it.

And THAT, ladies and gentlemen, is where the big problem occurs. The "decrypting" of the LOSSY compressed MP3 involves guessing what parts of the data is missing and attempting to re-create the original waveform (or more accurately, the original sample data). That conversion is mostly sufficient to "fool" the ear into thinking that the lossy compressed and that reproduced signal is "as good as" the original uncompressed signal. BUT trying to use this "re-constituted from juice concentrate" bitstream can make any subsequent editing steps (which are done in the UNCOMPRESSED domain) sound like steaming piles of fecal material. I know this from repeated first-hand experience (as well as the general consensus of everything I have read online since the birth of the internet)

HOWEVER, feel free to try it yourself. You don't have to take our word for it one way or the other. Fortunately in this case you can perform the experiment for yourself and draw your own conclusions.

 

[mhood]

BUT trying to use this "re-constituted from juice concentrate" bitstream can make any subsequent editing steps (which are done in the UNCOMPRESSED domain) sound like steaming piles of fecal material.

We called it "alien dolphin chatter". Edit with an MP3, compress it again for delivery and listen for it...you'll hear it. ;-)

 

[Noiz2]

Huh?

Unless I've been lied to for 20 years WAV files are sampled but NOT compressed. One may sample at a smaller word size and/or sample rate which will reduce file size (and fidelity) but the original captured fidelity remains within the confines of your chosen sample rate/size. An original 320 Kbps stream rate 16-bit 48kHz MP3, by it's very nature, will have less information than an original 16-bit 48Khz WAV of the same source. MP3 encoders, even at the highest stream rates, will always throw away data while applying "perceptual" encoding - trying to trick the ear into hearing something that's not there. MP3s are the JPEG of audio. JPEGs can look good but they're never as good as the RAW image captured from a camera sensor or original digital art work, and when you go to manipulate them they can cause big problems.

Fidelity, or how close the copy is to the original, in WAV files is a mainly a function of word size and sampling rate (ignoring the influence of the converters, etc.). No digital copy of an analog signal can be "perfect" but an original 24bit/96Khz WAV can come mighty close and you can mash it and bend it in post and it will still hold up. MP3s? Not so much.

Richard Thompson
T4 Media

It depends on how you define "compression". WAV files have no where close to the amount of "data" that the actual acoustic sound had. They reduce the data by only looking so often and only recording a given amount of information, the rest is thrown away. Since "compression" as used to talk about MP3's is describing a scheme to reduce the data required to store a sound, fundamentally WAV files are "compressed".

The "compression" used in MP3 and ATRAC etc. look at the data and give more storage to changing data and less to redundant data, there fore they are spending their bit budget much smarter than a brute force system. IF you encoded from the source sound using a smart compression rather than a bru=te force data reduction you would have a better representation of the original.

It's kind of moot because nobody is doing this. Your comparison with JPEG fails for the same reason comparing MP3 to wave fails. Today everyone is "compressing" the compressed recording. RAW is not the full image it's just the most that that particular sensor could capture. If you take that and then drastically reduce the data you will get inferior reproduction. If you used the JPEG codec and "compressed" to a file of the same size as the RAW file then you would only have loss due to math errors but you would not gain anything either. Though you might actually have a file that held up to processing better since it is capable of holding a lot more information than a RAW file could.

If you flip that though and fix your file size and then use a sensor that maxes out the storage capability of the format then your RAW file is still at the same resolution as before but your JPEG is now at a many times higher resolution.

These algorithms have mostly been used to either shrink a file for quicker distribution or increase the storage of consumer recorders. If you applied them instead to pack more resolution into the same size files you could have a major step up in acquisition quality.

So take your example Instead of a 24/96 WAV, you have a 48bit/480 khz recording, now which is going to be more "mashable"?

 

[Noiz2]

Yes, they can, as long as that MPEG compression is the LAST step in distribution. If you intend to do any post-production editing, then compressed files can be quite HORRID and are to be avoided if at all possible.

While for common usage I would agree, but you are referring to my post which was not talking about common usage so then we need to look at actual practice. A LOT of production dialog is recorded with wireless mics and all but the digital ones use companding which is just analog compression, and it's lossy. A lot of radio work is/was done on MiniDisk which uses compression. I said "compressed formats" not MPEG and there are a number of lossless "compressed" audio formats.

But yes compressed, expanded, tinkered with and re-compressed files can sound HORRID. Wiles that were never compressed can also sound HORRID. My point is that just because they can does not mean they have to.

That is simply not true at all. At least by the conventional definition of "compressed" as we use it in the audio world.

I explained in another post but you have your choice of definitions. Compression as a dynamic range modification or compression as a way to reduce data. A WAV file is a drastic reduction in the "data" of an acoustic sound. You could argue I guess that the later compression only applies to data in a computer being reduced but then most modern recorders have higher resolution A/D converters that the format that is recorded so you have that "in computer" reduction going on also.

"1 bit" is only one of several analog-to-digital conversion (ADC) or digital-to-analog conversion (DAC) techniques. It is more properly referred to as "delta-sigma" technique. "1-bit" is largely a marketing term and has no significant superior benefits compared to other ADC or DAC methods.

I'm not referring to the conversion but the recordings. There is a system used by KORG that is close to a format that was used for awhile for HD CD's. It's a significantly different way of recording digitally. And by all the accounts I have heard is pretty amazing.

I could not disagree with this statement more. EVERY lossy compressed bitstream (such as MP3) starts out as a "WAV file". Where I mean an original, unmolested stream of samples of the original waveform.

Lossy audio formats (such as MPEG, etc.) MIGHT sound virtually identical to the casual listener, but if you actually compare the signal (such as by inversion and cancellation) you will hear (and see if you are looking at an oscilloscope) the differences between lossy compression and the original signal.

I explained in another post. You are correct, but that is not what I was talking about.

True.

No. ALL digital
...
"magic" and has made perhaps billions of $£€ selling the implementation of those secrets.

Again you are correct and I was obviously not clear about what I was talking about and explained in a later post.

And THAT, ladies and gentlemen, is where the big problem occurs. The "decrypting" of the LOSSY compressed MP3 involves guessing what parts of the data is missing and attempting to re-create the original waveform (or more accurately, the original sample data). That conversion is mostly sufficient to "fool" the ear into thinking that the lossy compressed and that reproduced signal is "as good as" the original uncompressed signal. BUT trying to use this "re-constituted from juice concentrate" bitstream can make any subsequent editing steps (which are done in the UNCOMPRESSED domain) sound like steaming piles of fecal material. I know this from repeated first-hand experience (as well as the general consensus of everything I have read online since the birth of the internet)

HOWEVER, feel free to try it yourself. You don't have to take our word for it one way or the other. Fortunately in this case you can perform the experiment for yourself and draw your own conclusions.

I would disagree with the blanket nature of the statement. It certainly can be a problem. WAV files also guess to fil in the holes when ever you do any processing and the converters guess when you play it back. It is quite possible to record in a compressed format and convert it and work on it and not have a bunch of problems. It's very dependent on what you are recording and how you are recording it.

But as a rule talking about what is available today I agree. I would not by choice record into a compressed (lossy) format today. What everyone is jumping on was not a recommendation to do that. It was a bit of a rant on the backward nature that the recording equipment manufacturers have been working with compression and this completely false "truth" that a digital "uncompressed" file is a full reflection of the original.

It was also a reaction to the blanket knee jerk "truth" that your screwed if you have to use a compressed format. People in sound are very fond of the absolute rule. The trouble is that a lot of absolutes are not. From a teaching stand point I find that it causes another problem. One akin to the old government posters "Drugs Kill". Well except the ones that save your life, and how come you don't include nicotine or alcohol in there as "drugs"? If you go around saying absolutes and then someone makes the mistake and what you said always happens doesn't, they don't believe you about the stuff you are right about.

 

[Noiz2]

We called it "alien dolphin chatter". Edit with an MP3, compress it again for delivery and listen for it...you'll hear it. ;-)

Yes I have heard that also. Here is the catch though, what MP3? a high bit rate 48/24 MP3 or some massively compressed low bit rate junk, or even better a really low recording at a low bit rate that will get you there fast.


But take a 44.1/16 DAT transfered to WAV with older converters (like a lot of sounds in libraries) and start playing with that. You can get into digital gravel really quick, though not usually the artifacting you are talking about.

Garbage in, garbage out will always be a truth that holds up pretty well.

The other "classic" way to get there is to use Digi's BNR noise reduction on a 16 bit file aggressively. You can get everything from the alien voices to a nice underwater sound. I know folks who keep that plugin around just to create that effect.

 

[Richard Crowley]

Yes I have heard that also. Here is the catch though, what MP3? a high bit rate 48/24 MP3 or some massively compressed low bit rate junk, or even better a really low recording at a low bit rate that will get you there fast.

Even the highest rate MP3 compression (i.e. the LOWEST lossy data compression) will demonstrate the effect. That is why MP3 is NEVER a good choice for production.

But take a 44.1/16 DAT transfered to WAV with older converters

.
There is NO "conversion" involved in taking the (already digitized) audio data from the DAT tape and dumping it into an uncompressed container file (such as WAV or AIFF). Any more than taking the digital data from a hard drive or a flash memory device (such as a CF or SD card, etc. etc.)
The ones and zeros on the tape are transferred EXACTLY THE SAME, BIT-FOR BIT into the ones and zeroes in the uncompressed file. In fact, helical scan 4mm tape (i.e. "DAT") was used for several years as a backup format for computer hard drives. Digital tape stores ones and zeroes. Whether the bits describe an audio stream or video frames or bank account data is immaterial at this level.

Garbage in, garbage out will always be a truth that holds up pretty well.

Agreed. And ANY form of lossy-compressed audio codec (most notably MP3) is GARBAGE when it comes to audio production. Lossy compression is a DISTRIBUTION scheme used for space conservation. It is NEVER an appropriate selection for production.

 

[Noiz2]

Even the highest rate MP3 compression (i.e. the LOWEST lossy data compression) will demonstrate the effect. That is why MP3 is NEVER a good choice for production.

You mean the highest rate you have encoded. And truthfully most of the encoded field recordings I have dealt with are ATRAC not MP3 so maybe you can't get a decent recording in MP3, but I suspect you can.

But again my point was NOT about how MP3 is used today but how it could be used to allow much more data about the original sound. You insist at comparing apples to oranges. The highest MP3 data rate you can easily get your hands on is no where near the data rate of a WAV file.

Actually I'm also talking a bit of apples and oranges since I'm talking about recording at a much higher bit depth and much higher sample rate.

At the same bit depth and sample rate any lossy scheme will be inferior to some extent that a lossless scheme. There are however lossy schemes today that

There is NO "conversion" involved in taking the (already digitized) audio data from the DAT tape and dumping it into an uncompressed container file (such as WAV or AIFF). Any more than taking the digital data from a hard drive or a flash memory device (such as a CF or SD card, etc. etc.)
The ones and zeros on the tape are transferred EXACTLY THE SAME, BIT-FOR BIT into the ones and zeroes in the uncompressed file. In fact, helical scan 4mm tape (i.e. "DAT") was used for several years as a backup format for computer hard drives. Digital tape stores ones and zeroes. Whether the bits describe an audio stream or video frames or bank account data is immaterial at this level.

Ah, yes. It was common for DAT's to be transfered analog (in post) but of course you are correct. The A/D conversion would be happening only in the DAT if you did a digital transfer, which was common by the end of the DAT era. So I should have said take an old DAT 48/16 recording and mess with it and you will start getting junk also.

We are kind of dancing around an agreement. I agree that uncompressed is a more robust format. But it's not above turning to junk. It may take a bit more work to turn to junk but with old 16 bit recordings it didn't take much.

There may also be a bit of a music/ film mis match going on since a couple of folks talked about going back to MP3. In film post you are not going back to MP3 even if you started that way originally. You would be converting to a WAV file or some other non compressed format. You would never be working on the MP3 and you would never be converting it back to an MP3. You can have with compressed originals a number of issues. The one that the other poster called alien voices happens when the compression engine gets fooled by the incoming signal and starts putting out bogus data. I haven't seen it on the decode side but doppler tones on the encode side can cause this. Another is if you record low the codec has a hard time working the data and you can get "silence" artifacts where it's basically trying to encode it's self noise, also problematic.

Agreed. And ANY form of lossy-compressed audio codec (most notably MP3) is GARBAGE when it comes to audio production. Lossy compression is a DISTRIBUTION scheme used for space conservation. It is NEVER an appropriate selection for production.

So all wireless lav recordings are "GARBAGE"?

Blanket statements like that are just plain wrong. That is your opinion but it's not actually supported by fact and actual usage.

If you said you don't like them and in your experience they have been junk, that would probably be more accurate.

If you said They can be junk and are a lot more likely to be junk than WAV files I would actually agree with you.

If you went on to say that lossy codecs are never the first choice for production, I would agree with you.

They are not however "always" "garbage" and they are in certain cases the best/only choice for production.

It is used to conserve space, which is why in certain situations it is appropriate for production.

but then I never argued it was a good production format.

 

[paulears]

For goodness sake - Wireless analogue radio uses DYNAMIC compression, not DATA compression. Please, let's not start to muddy the waters even more. This argument has been running since the first CD was produced. Historically, the gradual development of CD audio has led the way. There have always been two completing design elements. First is accuracy - the process of capturing real sound storing it then recreating it. Essentially an analogue process, that suddenly involved two conversions to digital and back. The issues were in the accuracy of the sampling system, where the sound was chopped up, measured, and then turned back to sound. It's a bit like the Start Trek fans endless discussions on the transporter. If somebody uses one, are they the original, or a copy? The Technics people I believe introduced the concept of 1bit recording. I remember it being explained by the use of buckets at the time. 16 bit meant 16 buckets, almost but not quite the same size, being used to generate the samples that would be measured and recorded as a number. If each bucket was slightly different, then when the results were recreated the actual end product would have errors because of the slightly different bucket size. So use one bucket and just do the measurements faster. Nice and simple for people to understand.

Recording to digital means throwing data away, so a curved waveform converted to discrete steps is not accurate. The question is if it's accurate enough to be transparent? Once we start to selectively discard further data that an algorithm designer believes is superfluous, then the quality is compromised further.

So a wav, at any bit depth or sample rate is compromised from the original. Audio designers only consider data compression to have taken place when recorded data is discarded in the belief, it can't be heard. It doesn't really matter if it's gentle or severe, it's clearly not as good. Can it be heard, and is it good enough? we've not even considered error correction? Inventing missing data as a repair job!

Personally, I'm very happy using 320K MP3 files from my computer, I'm also happy with MD recordings on my many MD machines. I am happy with the HHB MD500 machines I have for location sound. Bringing these files into my computer systems rarely causes problems. Using somebody else's compressed recordings often reveals nasty artefacts - but the weird warblings mentioned earlier in the topic are not evident on the material produced on my MD500s. However, I record now with 24bit depth, in 44.1,48 and 96K sample rates, depending on what I'm doing. Converting from any format to another creates something less good, but I really do not have fast and hard rules.

Dynamic compression changes the sound, data compression can mangle it in different ways. My colleagues who work in broadcast often complain about jitter, and often state this as one of the more destructive problems in audio - spending lots of time and effort synchronising their audio paths in the quest for sonic accuracy. The average consumer seems to accept almost anything as ok, as long as it doesn't actual have silent gaps! I produced by accident one project in mono, and nobody even noticed!

 

[John Willett]

Recording to digital means throwing data away, so a curved waveform converted to discrete steps is not accurate. The question is if it's accurate enough to be transparent? Once we start to selectively discard further data that an algorithm designer believes is superfluous, then the quality is compromised further.

Paul, I agree with most of what you say - but not this bit.

PCM digital recording can perfectly recreate the original as long as the sampling freqency is more than double the highest frequency - IE: which is why 44.1kHz or 48kHz is used for an audio frequency response up to 20kHz.

There is *no* data thrown away at all - *nothing* is lost at all.

Even a 16-bit / 44.1kHz digital recording is far more accurate than any analogue recording was and no information is lost at all.

Saying what you do means that you do not fully understand digital recording - sorry.

I think you need to read up on your digital theory here - the John Watkinson book "The Art of Digital Audio" is a good place to start.


I do agree that any form of lossy recording - even MP3 at its best - should *never* be used for origination in any way at all.

 

[Richard Crowley]

There is *no* data thrown away at all - *nothing* is lost at all.

Even a 16-bit / 44.1kHz digital recording is far more accurate than any analogue recording was and no information is lost at all.

Absolutely. Saying that common "PCM" is "throwing away" data is completely contrary to the commonly accepted definition of ordinary digital audio.

To be sure, we can't take so many samples that EVERY point on the waveform curve is covered. But then when it comes right down to that, we can't do that in analog (magnetic tape), either. The waveform is "sampled' by the individual microscopic magnetic dipoles in the tape oxide. The linear speed of the tape determines the "sampling rate". And the random orientation of the dipoles contributes to random noise (aka. "tape hiss").

We can show theoretically/mathematically that sampling at a sufficient rate results in reproduction waveforms that have sufficient accuracy ("fidelity") to be indistinguishable from the original waveform. This work was developed over 50 years ago by Nyquist and Shannon and others at Bell Labs in the earliest days of digital audio research. This is why saying that linear PCM "throws away" data because of sampling is not an accepted view by most people. http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

Even with recording equipment and methods that record straight to compressed formats like MP3, the process BEGINS with linear PCM sampling and THEN the linear information is subsequently DATA REDUCED by the compression scheme: AAC, ATRAC, MP3, FLAC, Ogg Orbix, WMA, etc.

There are NON-lossy audio compression schemes as well as LOSSY compression schemes. Because of the way the LOSSY compression schemes work, renders them less than suitable for most audio production which involves further manipulation in post-production editing steps.

 

[paulears]

No John, I'm sorry - but in this example, the digital waveform is not an accurate reproduction of the original waveform. NO digital conversion can reproduce a sine wave. They get close enough for the audio to sound the same, but any sampling system has quantisation errors, and even if small, then the new version is NOT identical to the original, and some content has been lost. I'll make it clear I am not talking about the digital waveform having content reduced, I am talking about the data simply not being an accurate representation of the original. Any sampling system that samples in time ignores, or mis measures, or throws away (whatever you wish to call it) part of the original audio waveform. John Wilkinson's book does not contradict this whatsoever - my point, which is actually reinforced in John's book is that sampling is an inaccurate method of recording, storing and replaying audio.

So I have to disagree with John and Richard. If the digitally sampled data stream is not an accurate mirror of the analogue original, then what exactly would the difference be? Clearly there are sampling errors, even when sample rates and bit depths are increased to the current very high levels. The digital version is inferior, parts of the original are not present. The data has errors in the stream and is 'less' than the original.

Please explain to me how a sampled waveform can ever be anything other than nearly as good - something is missing. Even if the errors are very small and totally inaudible, they are still errors. I was very clear that this was not a form of data reduction - clearly a different topic, but I stand by my opinion that things are missing.

If this is NOT the case, then why would we be sampling at higher speeds, to reduce what? If recording at 16/44.1 is accurate, why would we want 24/96? Because the resultant data is more accurate, but this then produces the conclusion that the lower data rate had something missing and was inaccurate ....... wouldn't it.

I'm surprised at your tone, John - ticking me off like that wasn't really required, was it - especially as you (hopefully) misunderstood my comments. Please do tell me if any of this is wrong? I'd love to be put right - but as I have patiently attempted to explain, I cannot accept that I am wrong in commenting that not all non-data reduced streams are equal. Sampling quantisation errors do mean that something has been lost.

I'm quite comfortable with Nyquist - but we're using band limited conclusions - in this case related to accuracy in a specified band. This got great coverage when CDs first popped up, so people concluded that the audio was accurate in conversion. Now we realise that perhaps this was a little er, presumptive, so we push the spec a bit and hopefully agree to ignore errors outside of what we can hear. All good stuff for discussion!

 

[Richard Crowley]

Please explain to me how a sampled waveform can ever be anything other than nearly as good - something is missing.

By that method then, there is NO method of recording sound (or picture, for that matter) that is perfect.
For that matter, even your own ears (and eyes) introduce errors from what the true audible/visible signal is while on the fly.
You can split hairs here, but you (and Scott) have rather distracted from the original topic. If you want to argue how good digital audio is, go for it. I'm not particularly interested in that pissing match.
But the original question was LOSSY DATA COMPRESSED vs. UNCOMPRESSED digital audio. And there can hardly be any controversy over that.

 

[Alex H.]

Alright, folks.

Let's get back on topic without the heated argument.

 

[Noiz2]

For goodness sake - Wireless analogue radio uses DYNAMIC compression, not DATA compression.

It uses companding, wich is analog "data" compression. Maybe we should start using the term "information" instead of data.

So more accurately digital recordings throw away information.

Recording to digital means throwing data away, so a curved waveform converted to discrete steps is not accurate. The question is if it's accurate enough to be transparent? Once we start to selectively discard further data that an algorithm designer believes is superfluous, then the quality is compromised further.

The average consumer seems to accept almost anything as ok, as long as it doesn't actual have silent gaps! I produced by accident one project in mono, and nobody even noticed!

The folks who think all compressed formats are garbage, have to tack on that they are only garbage if you use them for acquisition because every film you have heard in quite a while that wasn't in stereo was basically MP3 ,it's a Motion Pictures format and is essentially what Dolby Digital uses.