99% Of Our Music Is Lossy

[The Computer Audiophile]

By my estimation based solely on experience 99% of music we purchase is lossy. Unless we receive master file copies, like Reference Recordings HRx 24/176.4 or Barry Diament's Soundkeeper albums at 24/192, we are receiving a lossy version. Most music isn't recorded at 16/44.1 but we've been willing to call it lossless. Where should the lossless / lossy line be drawn and why?

[Julf]

By my estimation based solely on experience 99% of music we purchase is lossy. Unless we receive master file copies, like Reference Recordings HRx 24/176.4 or Barry Diament's Soundkeeper albums at 24/192, we are receiving a lossy version. Most music isn't recorded at 16/44.1 but we've been willing to call it lossless. Where should the lossless / lossy line be drawn and why?

 

Well, any music we purchase is lossy - inevitably some information gets lost in the recording process between the sound of the instrument and/or artist and the recording. The terms "lossy" or "lossless" don't really apply to a file or a recording, they apply to a process that the music information is subjected to - be it encoding/decoding, transmission over a network, or some other form of processing. My understanding is that the vast majority of music today is recorded at 24/48 - all we can hope for is to take that data and reproduce it with as little additional (audible) information loss as possible.

[MikeJazz]

Most music isn't recorded at 16/44.1 but we've been willing to call it lossless.

 

Your point is that only if we receive the same resolutions as in the master we have a lossless copy...

OK...I guess the industry come to an agreement that CD 16/44 would be enough for a complete experience.

 

Also because off this I always tell me friends that buying iTunes mp3, is really a bad business. As the format is just a (lesser) copy of the original, a worst copy always and does not full justice to the work.

 

However, I think we should focus more on the recording/engineering/mastering quality. And to promote the use of good engineers, because gradually they are being either discarded either to cut costs or to let bands be more "In control"...

 

I prefer to listen to a good mastered mp3 320 lossy file, than a bad mastered 24/192 master file or CD.

I had been enjoying linn radio at that lossy resolution and many tracks sound way better that many CD's...

[Jud]

My understanding is that the vast majority of music today is recorded at 24/48....

 

Interesting. Source?

[The Computer Audiophile]

Well, any music we purchase is lossy - inevitably some information gets lost in the recording process between the sound of the instrument and/or artist and the recording. The terms "lossy" or "lossless" don't really apply to a file or a recording, they apply to a process that the music information is subjected to - be it encoding/decoding, transmission over a network, or some other form of processing. My understanding is that the vast majority of music today is recorded at 24/48 - all we can hope for is to take that data and reproduce it with as little additional (audible) information loss as possible.

 

Hi Julf - I'm only talking about the process the music is subjected to, not the loss from real performance to capture. If music is recorded at 24/48 uncompressed and we receive 16/44.1 uncompressed we're really receiving a lossy copy that people are willing to call lossless.

 

I'll go back to my original question as I'm interested in your opinion.

 

Where should the lossless / lossy line be drawn and why?

[Audio_ELF]

Hi Julf - I'm only talking about the process the music is subjected to, not the loss from real performance to capture. If music is recorded at 24/48 uncompressed and we receive 16/44.1 uncompressed we're really receiving a lossy copy that people are willing to call lossless.

 

I'll go back to my original question as I'm interested in your opinion.

 

Where should the lossless / lossy line be drawn and why?

Lossless or lossy really doesn't have a meaning in this context (IMO).

 

What would be nice is if we could expect some kind of workflow details so we can make our mind up - a bit like the AAD / DDD markings on CDs of old but more detailed. It would also help with the "up sampled 16/44.1 sold as 24/192 debacle).

 

Eloise

[Julf]

Where should the lossless / lossy line be drawn and why?

 

As I tried to state, and Eloise stated better than I did, "lossless" or "lossy" only applies to an encoding or process, not a recording. 16/44.1 FLAC is a lossless storage format, but if you take a 24/48 recording and downsample it to 16/44.1`to be able to make a 16/44.1 FLAC out of it, that process is of course not lossless.

 

So my opinion is that "lossy" vs "lossless" onlu makes sense when we talk about processes and codecs. When we talk about recordings, we need., as Eloise says, agreed labels such as "original resolution", "CD quality", "upsampled" etc.

[The Computer Audiophile]

Lossless or lossy really doesn't have a meaning in this context (IMO).

 

What would be nice is if we could expect some kind of workflow details so we can make our mind up - a bit like the AAD / DDD markings on CDs of old but more detailed. It would also help with the "up sampled 16/44.1 sold as 24/192 debacle).

 

Eloise

 

Hi Eloise - I would also love more information about recordings from start to finish.

 

Why don't you think lossless or lossy apply to this situation? I'm really interested in people's opinions on this.

 

I think it applies because a label releasing 256kbps through iTunes and 1411kbps through physical media when the original is something like 4608kbps. I think there are many more variables involved in sound quality, but lossy and lossless don't address sound quality directly.

[Julf]

My understanding is that the vast majority of music today is recorded at 24/48A

Interesting. Source?

 

Just discussions with people in the industry, and comments I have seen elsewhere. Hopefully some of the CA members associated with the recording industry can provide more solid data.

[Julf]

However, I think we should focus more on the recording/engineering/mastering quality. And to promote the use of good engineers, because gradually they are being either discarded either to cut costs or to let bands be more "In control"...

 

I agree 100%...

[The Computer Audiophile]

 

 

Just discussions with people in the industry, and comments I have seen elsewhere. Hopefully some of the CA members associated with the recording industry can provide more solid data.

 

My understanding is that 88.2 followed by 96 k are more popular. No real evidence other than word of mouth :~)

[bdiament]

Hi Julf,

 

...My understanding is that the vast majority of music today is recorded at 24/48...

 

Actually, the overwhelming majority of what I see recorded today, in both pro studios and semi-pro, tends to be 24/44.1.

 

I've spoken with folks about using higher rates and the common theme so far has been that given the typical number of tracks in use today (and the most common converters), their systems "choke" with higher sample rates.

 

A *very* tiny minority have systems capable of doing multichannel recordings at higher rates.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[audiozorro]

The lossless / lossy line should be 24 bits. Nowadays, is there any reason to record at less than 24 bits? As far as sample rates, I prefer 176 kHz or DSD64 and higher for stereo but I am willing to accept 88.2 or 96 kHz for multichannel audio.

 

True this line is arbitrary and for many meaningless, but advances in computer audio, digital transmission and digital storage allows the bar to be raised and free us from the limitations of old formats and standards. Perhaps the more meaningful standard would be a fractional comparison of the sonic qualities between the released album and the highest studio master that the label possesses (the Measurement Junkies should love this one).

 

More and more labels will jump on the high rez digital bandwagon and offer us album releases with greater resolution, dynamics and clarity. Even older analog recordings (there’s gold in them hills) benefit as the decaying master tapes and records are preserved in DSD64 and DSD128 formats. We should all commend those labels willing to release their digital studio master files at affordable prices.

 

And true, 24/88.2 and higher resolutions does not guarantee better sound but it is generally accepted that the studio masters represent the best recordings. And I suppose where a studio master is not up to the highest musical standards of high fidelity, there are several options for improvement including remastering, enhancing, removing flaws or as some ingenious labels have successfully demonstrated creating impressive music events that never actually occurred in history.

 

I can’t say with authority how the music industry is doing, but I have never had so much great music available at my fingertips (a King’s treasure trove of music) and I have just scratched the surface.

[The Computer Audiophile]

As this is my thread and created for the specific purpose of discussing the following question, please don't drag it into the abyss thats already been discussed countless times. Further off topic or semi off topic posts will be removed.

 

Where should the lossless / lossy line be drawn and why?

[Jud]

As this is my thread and created for the specific purpose of discussing the following question, please don't drag it into the abyss thats already been discussed countless times. Further off topic or semi off topic posts will be removed.

 

Where should the lossless / lossy line be drawn and why?

 

(Grits teeth with effort) "Must -pull back -from abyss!"

 

My own thinking is that the line has already been drawn for us by common terminology. Lossy/lossless are terms commonly used for what's done to a file subsequent to the recording process (without right now getting deeper into the details of whether we should only be using such terms in connection with compression). The terms commonly used for what's done during the recording process are, for better or worse, more along the lines of "hi-res," "DSD," "DSD64," "DSD128," "DXD," etc.

[RobbieC]

Sorry Chris. Took too long to write my reply to see your admonishment.

[The Computer Audiophile]

Sorry Chris. Took too long to write my reply to see your admonishment.

 

No worries. I moved some posts into a separate thread.

[Paul R]

Such a good question.

 

I probably have an outlier opinion on this, but I think the original recording specs should be specified, and any distribution with any specifications lower than that spec should be considered "lossy."

 

-Paul

[audiozorro]

Doesn’t a lossless format refer to its ability to recreate the original data exactly? Thus if the original recording data is a 24 bit studio master file, how could any 16 bit or lower file be anything but lossy?

 

Of course if one assumes that the 16 bit/44.1kHz CD is the original data then you and the labels have artificially lowered the requirements for lossless audio. And hence it is easy to state that 99% of the music in consumer hands is lossy.

 

Perhaps someone can elaborate on the technical details of converting a 24 bit audio file to a 16 bit file and discuss what is lost.

 

Conversely I would assume that upsampling or converting an original 16 bit file to a 24 bit audio file can always be reversed to recreate the original data?

 

Personally I am satisfied with those labels that release music in various formats from low rez lossy files to high rez studio masters to give consumers a choice.

[RobbieC]

I probably have an outlier opinion on this, but I think the original recording specs should be specified, and any distribution with any specifications lower than that spec should be considered "lossy."

 

-Paul

 

I really like this idea the best.

[ecwl]

I don't know the history of lossy vs lossless in music files. But I always thought it came from photos. We would never call a GIF or TIFF lossy but JPG is lossy. It's not that a TIFF cannot be a downsampled version of the original file, it's just that the JPG is a perceptually lossy compression of the original file.

 

I think that's where we should make the distinction. 16/44 may be downsampled from higher resolution files just as most photos on the net are downsampled files of larger megapixel files. But mp3 actually performs lossy perceptual removal of sound information just as jpg does to photos.

 

So I'd argue most of our music is downsampled but not lossy.

[RobbieC]

Perhaps someone can elaborate on the technical details of converting a 24 bit audio file to a 16 bit file and discuss what is lost.

 

I can...but not in this thread as requested by the OP. Briefly, dropping from 24 bit to 16 bit can (1) raise the noise floor causing a loss of very, very low amplitude components of the signal and/or (2) increase a type of correlated distortion called non-linearity distortion. It's only slightly technical.

[bdiament]

My take is that though I (believe I) know what you mean Chris, using the term "lossy" in this context is as mistaken as using the term "lossless" to describe PCM formats (as used by anyone I know in record making who uses them). Those terms certainly apply (at least, in some cases, in theory) to data reduction codecs.

 

Maybe what is needed, as you and others have said, is some mention by the record label of the original recording format. I know this is simply not possible in the case of many older recordings, where the information was never written down and the best we can hope for is someone that was there.

 

We live in a new audio age where, for the first time, what the listener *can* get at home is not necessarily a reduction from what was created in the studio, as it has always been in the past.

 

Of course, knowing something was recorded at say, 24/192 is ultimately still not enough. For myself, information about the microphones used, how many and how they were placed (and why) tells me a whole lot more. My experience has been that 90-95% of a recording's ultimate quality has already been determined by the time the signals are leaving the microphones. In many, if not most ways, *that* is where the largest losses occur. I've said before, I hear more "Life" in an mp3 sourced from a Keith Johnson recording than in the original masters made by many other engineers.

 

But I understand this thread is about the original *capture* format and how what we purchase compares with that. In that sense, much of what we can buy has already suffered a degradation. Still, if most of the multitracks I've encountered were done at 24/44, where does this leave us? For my ears, the best 96k I've heard *still* loses to the direct mic feed. A lot of 192k does too since not that many converters have the clocking and analog stages to get the full potential of the format, regardless of what the spec sheets and reviewers claim.

 

The best we can get from that 24/44 multitrack original is a 24/44 file to play at home. (Yes, it can be passed through analog and reconverted at a higher rate or it can simply be unconverted but the ceiling has already been set.) In the end, the original digital capture format (i.e., word length and sample rate) is what I'd prefer.

 

Hopefully, we'll see great use of high resolution by more recordists as their system horsepower grows sufficiently to accommodate dozens of tracks at high sample rates.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Alkan]

 

However, I think we should focus more on the recording/engineering/mastering quality. And to promote the use of good engineers, because gradually they are being either discarded either to cut costs or to let bands be more "In control"...

 

I prefer to listen to a good mastered mp3 320 lossy file, than a bad mastered 24/192 master file or CD....

 

I agree 200% !

When I think to have a musical lossless experience, I think in something closer to the live experience. Unfortunately, as I can hear in abundance, the engineering / mixing / mastering process has changed to "loud" albuns (besides problems with stage position, depth, airy, transparency etc.)

The worst DR album from my JRiver library is 24/48 - like the 2nd worst.

[mwheelerk]

Chris

 

Let's say your basic premise is correct. As I understand it then anything less than the original recording would be lossy. I have always thought of lossy and lossless as pertaining to the method used by audio codecs in converting PCM to other portable file formats ( our WAV, FLAC, ALAC, mp3 etc). So if we already have primarily lossy what do we then have when audio codecs are applied? Lossy lossy? Lossy Lossless?

 

I am far from any expert but my personal understanding would use it in regards to the application of an audio codec to an original source (which may or may not be lossy according to my understanding of your premise).