Is It Time To Rethink Lossless?

    Audio: Listen to this article.

I never thought I’d title an article, Is It Time To Rethink Lossless. I’ve always understood the definition of lossless and found solace in the fact that everyone from the audiophile community to the more mainstream audio community and from the subjective to the objective communities, spoke a common language with respect to lossless. Now it seems the term lossless is being stretched and twisted and applied to situations where it may not be indicative of what people think when they hear lossless. Let me explain. 

Lossless compression has always meant that an audio file is compressed from its original size to something smaller, and at a later time can be uncompressed to the exact same original file. Nothing is lost. 

A CD is delivered to the consumer, the consumer rips the CD to FLAC, WAV, ALAC, etc… Notice I had to throw WAV in the mix. That’s lossless as well, but uncompressed. It’s lossless in the sense that nothing is lost from the original CD. FLAC and ALAC files could be uncompressed to the identical files on the CDs, without loss of any music. For the sake of this discussion, let’s stick to lossless in the compression sense. 

Quoting a rather dry Wikipedia article:

“Lossless compression is a class of data compression that allows the original data to be perfectly reconstructed from the compressed data with no loss of information. Lossless compression is possible because most real-world data exhibits statistical redundancy. By contrast, lossy compression permits reconstruction only of an approximation of the original data, though usually with greatly improved compression rates (and therefore reduced media sizes).”

Why Rethink This Basic Concept?

It may be time to rethink how we use the term lossless. In order to be lossless, there has to be an original source and compressed version that could be turned back into the original source. That CD, turned into FLAC files, could be turned back into the identical CD. Sounds pretty pedestrian. 

Loosening up the usage of the term lossless started when high resolution audio was released. We no longer had a physical product to rip. We were presented with 24/96, 24/192, etc… files and we just called them lossless. In a way, we took the term for granted because we had no source to which we could compare. Sure, if we wanted to convert the high resolution download into something else, we would consider our download the source and turn it into FLAC or ALAC etc… But in a way, that’s like getting an MP3 from Napster and calling it one’s source because that’s one’s original file. I guess one could do that, but it isn’t the wisest move. 

Thinking more about high resolution files / streams and the original source, we would be crazy to think that all of this high resolution audio is a lossless version of the original source. The works both ways up and down the sample rate scale. For example, NativeDSD sells albums at 24/96, 24/192, 24/384, DSD, DSD128, DSD256, DSD512, etc… Some times all these resolutions are available for the same album. I’m not complaining or singling NativeDSD out for doing anything negative whatsoever. The store is just an example that comes to mind because I frequently browse the site. Anyway, it’s incredibly likely that only a single one of the available rates is lossless to the original. Yet, we don’t think of calling the others lossy. At least I don’t think of doing that, but should I? Probably not, but perhaps a more nuanced description is needed. 

On the other hand we have files and streams at 16/44.1, 24/48, 24/96 and 24/192 from almost every record label. The likelihood that these are truly lossless to the original source is, unknown. It’s likely unknown to the artist, the label, and most everyone involved in creating the album. We can set aside the digital watermarking in streaming as that’s an entire different can of worms. But, we should think about the murkiness of calling this music lossless when we don’t really have an idea if it’s true. We know it’s delivered in a lossless container, but then again, so was MQA. 

Many professionals use higher sample rates when working with music in their digital audio workstations, then output to a lower resolution delivery format. The extra headroom is seen as necessary for the processing, but unnecessary for delivery/playback. Think of all the CDs and 16/44.1 albums available for streaming. Many originated at, god forbid, 24/48 before being downsampled for delivery. Yet, my CDs were lossless dang it. 

Now we venture into immersive audio, where the highest quality files ever released by most labels are, to date, 24/48 768 kbps Dolby Digital Plus Atmos (keep in mind that many of us have stereo FLAC files at 400-500 kbps, but that’s neither here nor there for this discussion). Dolby workstation tools accept up through 24/96 audio, and many audiophile labels work with music as high as 32 bit / 384 kHz before creating the Atmos version. 

Is it appropriate to call the streaming Atmos albums lossy, when they are the highest resolution ever released to the public? Are they lossy only because they were delivered in a lossy DD+ container? What about a downsampled CD delivered in a lossless container? Neither could be used to recreate the original, using a strict definition of lossless compression. 

Must we consider losslessness in the context of the original or the version released to the public? In other words, lossless compared to what? Discrete immersive audio, the holy grail of immersive music, is usually released at 24/352.8 in ten to twelve channels. These are output from a digital audio workstation, with some parameters, to WAV files and considered lossless by everyone. The same workstation usually outputs a 24/96 version to be used in creating the TrueHD Dolby Atmos version that’s delivered at 24/48. These albums are also considered lossless by everyone. 

The same workstation, using the same 24/96 Atmos master as above, also uploads the 24/48 ADMBWF Atmos Master to Apple. Apple then encodes is using Dolby Digital Plus, for delivery through Apple Music. For 99% of releases, this Dolby Digital Plus version is the highest resolution ever released. 

Working backward we can see neither the streaming Dolby Digital Plus 24/48 version, the TrueHD 24/48, nor the Atmos master at 24/96, could be used to recreate the original 24/352.8 files losslessly. Yet, we only consider the highest resolution released to the public as lossy. Using this logic, we should retroactively call most CDs lossy and most 16/44.1 streams lossy. We’d also need a lot more information about other albums, which could mean changing the definition of an album from lossless to lossy, even though nothing in the musical data has changed. It’s a bit crazy.  

Yes, We Should Rethink Lossless

The term lossless has always been seen as black and white. A simple concept used to label our favorite music. For the most part, subjective and objective audiophiles have all agreed on what’s lossless and what isn’t (save for the short period of time when some believed MQA was lossless). However, I believe we need to rethink usage of the term lossless. We should never alter the definition. After all, lossless to a specific source is a very real and valuable concept. We should rethink when and where we use the term lossless. Lossless to what, the one original master of the recording or a studio working version at a slightly different sample rate or the highest resolution released to the public or something else entirely? 

It’s even possible that the best use of the term only applies to situations in which we know everything about a recording or it’s very obvious. For example, Spotify delivers lossy OGG music. That’s easy. No matter what definition of lossless we use, I’ll bet the farm that 99.999% of the albums didn’t originate with a 320 kbps Ogg Vorbis master and we have higher resolution versions released in lossless containers available elsewhere. Another day one is vinyl. It’s lossy, unless it’s a direct to disc album. Period. After this, it gets really sticky, and inaccurate to apply the term lossless to all but one’s own format conversions and those rare albums about which we know everything. 

Lastly, I didn’t write this to be controversial. I honestly think we should rethink lossless and have a lively discussion about when and where we as music lovers use the term lossless. Lossless has a definition. Let’s use it accurately.