Quick mp3 question

[gsc1ugs]

another massive factor to me is compressed audio, that coldplay one is awwwwwwefull

[Sam Spoons]

Just a question, are you dissatisfied with the sound of your present .mp3 tracks? If they are commercial tracks they will most likely be high bitrate mp3s and IME the artefacts present in high bitrate .mp3s will not be noticed by the majority of listeners (those of us who have learned to recognise them are, obviously, a different matter) and I doubt most of your audience would notice (I've never had anybody complain about the quality of .mp3 sourced break music on a gig).

another massive factor to me is compressed audio, that coldplay one is awwwwwwefull

There's a fundamental difference between audio compression (as in that Coldplay track) and data compression (as in .mp3s).

[gsc1ugs]

Most are very good, i do have some under 128 but original question was about digital playback, i use my daily iPhone for my gigs its great just wanted clear up that playback argument but seem to have opened a can of worms. My provider is kindly sending me wav versions to eliminate that problem for small fee

[The Elf]

I dont have the masters they are sold as mp3 even paris music

A perennial problem when I'm working with artists who bring me backing tracks. Why backing providers seem reluctant to offer lossless audio is silly and annoying!

And, to echo others here... DON'T convert MP3 to WAV!!! Not only is it pointless, but should it be converted back again the quality goes down the toilet.

[Drew Stephenson]

My provider is kindly sending me wav versions to eliminate that problem for small fee

Sounds like the best solution.

[gsc1ugs]

Cheers folks

[Martin Walker]

if anyone is till interested in exploring the mid-side stuff in more detail, I've tracked down an old izotope blog from 2014 that first got me excited about the possibilities:

https://www.izotope.com/en/blog/mixing/ ... nking.html

Here's an idea of what's on offer:

"Listening In Side
I started thinking about this some years ago after I had the good fortune to serve as host for mastering engineer Bob Ludwig when he came to talk to my students at SUNY/Fredonia. In the course of some fairly extended conversation, he told me that he always listened to the Side version of recordings that he mastered and that he'd come to find them very interesting from a production standpoint as well as a technical one. He found various producers' characteristic practices and tricks were much more audible in Side, and that many elements of the original tracks were more clearly revealed in Side. He suggested that my students might find it useful to listen to recordings this way in order to quickly identify and comprehend various producers' styles."

And of course there's plenty on offer from the SOS archives too:

https://www.soundonsound.com/techniques ... processing

Martin

[AlecSp]

Late to the party on this one, but mp3 bitrates:

In my original ripping experience I was naive on the bitrate and didn't discover until later that I'd been using a default rate of 128kbps. Nothing emerged on the cheap PC earphones I was using with that. My shock came when I used a 128kbps copy of Bohemian Rhapsody at a funeral on a church's cheap PA - it was embarassing. Genuinely that bad.

This prompted some comparative listening, and 128 vs 256 kbps mp3s were night & day different. But I quickly moved to 320kbps as the storage increase was minimal. In all practical situations I am happy that the quality of a 320kbps mp3 is not the limitation in the signal chains that I use.

Sadly, despite what some people would like to think, we're in a world of MP3 for music consumption. Lossless formats may be "better", but they're not as universally usable. So, for me, utility outweighs the last smidgin of notional technical performance and tips the balance back in favour of (decently encoded) MP3

Don't forget that what you're listening to can never really sound better than the worst part of the signal chain.

[hobbyist]

I thought I saw results from AES many many years ago on topics like this, doing ABCX double blind tests.

There have been lots of proper double blind testing of mp3 and many other lossy codecs over the years, reported in depth by pretty much all the standards bodies. The audibility of coding artefacts is almost entirely dependent on the coding bit-rate. Most people would probably agree that low bit-rate mp3s sound terrible. Equally, most people would find it difficult to distinguish between high bit-rate mp3 and an original source wav...

I recall they saying that some folks could identify when they used 192 and 384 rates and said they sounded better. That was A/D/A not mp3 specifically though.

Maybe I'm misunderstanding here, but you appear to be confusing converter sample rates with lossy-codec bit rates... .

Storage is cheap now. There is no need to compress to save memory or disk space like the old days.

Perhaps not... And broadband rates are generally much higher too so it's less necessary for streaming services too... But it will be a long time before lossy codecs genuinely become unnecessary and obsolete.

I can tell the difference between mp3s and wav even though I am not one of the golden eared folks.

Not confusing. The article I read was about converters. I just extrapolated to saying AES must have done similar tests for various bit rates on codecs as well as other comparisons.

As far as I am concerned, lossy codecs are obsolete now. I agree that they will be around for a long time anyway.

[CS70]

I was listening via a bluetooth connection to my hifi

For the MP3s, yeah you may or may not have another layer of encoding on top - it depends on the bluetooth devices. The standard common-denominator LCSC encoder is indeed awful for music due to the narrow bandwidth, but a bluetooth source/receiver couple can support MP3 streaming directly if both source and receiver ("sink") agree to it.

For the WAV however, it will definitely be subject to lossy compression, so you're not listening to the actual PCM data over bluetooth. So what you were comparing were, at best, two different lossy codecs (and at worst, two equally bad LCSC-encoded audio streams)

The best is to simply access the sound file as a data storage (so it's sent verbatim) and play it over the monitors via an audio interface. Or, if you want to go hifi, either go cabled or burn a cd

[Hugh Robjohns]

As far as I am concerned, lossy codecs are obsolete now. I agree that they will be around for a long time anyway.

Just for clarity, obsolete means 'no longer produced or used' but, sadly, neither is the case for lossy-codecs in general, or MP3 specifically.

I agree with your (earlier) point that in many cases it is now perfectly practical to use raw linear PCM tracks, or encode with loss-less codecs -- because mass data storage is so cheap, and data bandwidth is now very high in many situations (cable/fibre broadband, 4G/5G mobile broadband etc).

So, like you, I would definitely urge people to use either raw linear PCM or loss-less codecs (FLAC, ALAC etc) whenever they possibly can.

Personally, I like to use ALAC because it is a bit-accurate loss-less encoder that saves around 50% of the raw files space, but it has all the convenient standardised metadata capabilities of MP3, which I find extremely useful. Flac is much less well organised in this respect...

However, the fact is that lossy codecs are very far from being 'obsolete', and it would be very ill-advised indeed to think they are.

Lossy codecs are employed in all the current Satellite and DTT TV broadcasting services, and in digital radio, for example. They are employed in the soundtracks of DVDs and many Blu-ray discs. They are also employed in all the streaming TV media services (Netflix, Amazon Prime etc, Youtube), as well as in music feeds like Spotify, Deezer etc.

As you say, none of these will be going away in the foreseeable future, and while some services do now offer a loss-less codec option (eg, Deezer Hi-fi) for a higher monthly fee the take up is not exactly huge...

So I'd say that while lossy-codecs are avoidable in many situations, they are far from obsolete.

H

[gsc1ugs]

In a nutshell

[gsc1ugs]

My wav files are in the post even if my ears cant hear 128bit v wav

[Kwackman]

In a nutshell

Nope.
The sample rate (should be 44.1 kHz) and word length can be different in either format.
A WAV file can be (and on audio CDs are) 44.1kHz, 16bit.

There may be other good articles in the SOS archive, but I found these.
Get a large cup of coffee.

https://www.soundonsound.com/sound-advi ... udio-files

https://www.soundonsound.com/techniques ... your-music

[CS70]

Sadly, despite what some people would like to think, we're in a world of MP3 for music consumption. Lossless formats may be "better", but they're not as universally usable. So, for me, utility outweighs the last smidgin of notional technical performance and tips the balance back in favour of (decently encoded) MP3

Don't forget that what you're listening to can never really sound better than the worst part of the signal chain.

The difficulty with these things is that it's easy to overgeneralize - the devil is often in the the details (i.e. the context of usage).

Lossy codecs don't just cut away random pieces of sound - the original work at Fraunhofer was the pinnacle of two decades of research by people who knew what they were doing and build up on existing science (general error-correcting coding for example) and invented some new to do achieve the compression - psychoacoustic coding and perceptual entropy, for example.

Sure, the science is based on certain assumptions but these assumptions (and the corresponding algorithms) have been tested, verified and adapted several times over years before the format was finalized. In comparison with the similar work at JPEG, most of the pictures we see on the webs or even in magazines are compressed and nobody is the wiser. Try to print a mega-poster from them tough, and you'll see odd stuff if you go near.. but a megaposter is supposed to be seen from afar, so even in that case JPEG is effective.

For a phone jingle, even a very low bitrate lossy codec work just fine. A 128Kpbs MP3 is not necessarily bad: the Germans' theory was that it could actually represent quite well a CD quality signal while compressing size down to an average of 12 times. Constant bitrate is simply sample rate for the bit depth, so16bit, 44.1Khz sampling rate, which for two channels means 1411.2kbps - going down to 128Kbps means 11 times lower, with the final step being normal Huffman compression. Every step effectiveness depends also on the material - even the simple data encoding really depends on how many sequences of bit repeat in the data stream, and for how long.

Here's how it works In brief (and cutting a lot of detail), for each frame (1152 samples), there's 32 sub-bands - in line with the way that most people can hear differences in the various frequencies. The actual width depends on the sampling rate, and each samples is chuncked down to the bands (i.e. multiple samples, containing their specific part of the spectrum, are created from one). Since it's filtering, some aliasing will be introduced. Then 31 of the 32 samples are discarded, i.e. only a specific frequency are is represented in that sample - all the other bits are zero. Then "windowing" is applied - to average the signal and basically reduce the impact of the sample boundaries - and it's a critical step, as the algorithm looks at how similar consecutive samples are and - if they are similar - uses a "long" window, if they are not, a "short" one: a long window effectively says "the signal stays the same", and does reduce a certain amount of detail. The trick is in the definition of "similar". The psychoacoustic model is applied during windowing to provide information for that - leading, if necessary, to discard parts which are unadible according to an assumed model of human hearing - it tells the algorithm when to use short windows or long windows. The result is further processed to scale the data so that noise remains below what the PA model say is imperceptible, and the the optimal quantization step size for Huffman encoding (to find the "words" which are meaningful to look at). Finally, the resulting samples sequences is Huffman-encoded (in other words, zipped ).

So the algorithm depends on how finely the average ear can distinguish bands, and the PA model determine what is considered discardable, what is perceived as "the same sound" for each band and what noise is considered imperceptible. And that's why certain people can hear stuff and other don't.

Also, the standard leaves quite a bit open to the implementation, so it really makes no sense to talk of "Mp3 encoding at XX bitrate" by itself. What encoder is used is important! For example, what do we do with freqs over 16KHz? Some encoders may discard them entirely - and a young person with good ears may notice it; some encoder won't and even develop their own models to deal with these freqs. For example Lame mp3 is a very good open source encoder - perhaps the best.. we can't know as others are closed-source.

Not to mention variable bitrate encoding, where other models are used to decided when to keep the bitrate low or increase it temporarily, by preprocessing and analyzing the data stream!

So the science has (as always) limits: it can tell what the averages are, but by definition different data streams may yield different result, and a large amount of empirical data is needed to get some statistical conclusion. After many years of adoption, it seems that with a good encoder, a 256/320kps bitrate is definitely good enough, and 128kps holds water in many cases for most people, as theorized. No surprise most people don't complain!

Exactly like any other tool, if time and cost permit, it's always best to use the best available - so, in days far removed from 1992 (i.e. abundant, cheap fast storage), lossless files are certainly the way to go. But as Hugh noticed, in limited bandwidth situation (as over a cellular network for example) the compressing power of MP3 can still be of great use.

[Hugh Robjohns]

The difficulty with these things is that it's easy to overgeneralize - the devil is often in the the details (i.e. the context of usage).

Absolutely. Complaining about (or praising) MP3 is meaningless without also specifying the bit-rate (and to a lesser extent, the coding scheme) being used. The different coder designs aren't all equal either (see below), so you really should identify the specific coder being used, too.

High bit-rate MP3s are very hard to distinguish from the linear PCM source, by design. The proportion of people that can reliably detect artefacts is intentionally -- and demonstrably -- very, very small. The peer-reviewed research has been done and confirmed.

Conversely, low bit-rate MP3s tend to have more obvious artefacts that many will hear quite easily -- although a lot depends on the nature of the source material and how well it can mask the coding artefacts. A well-recorded talking book (single male speech) recording, for example, encoded at 128kbps may well sound perfectly acceptable and artefact-free, whereas a dynamic orchestral concerto at the same rate will probably sound pretty poor to many people.

Sure, the science is based on certain assumptions but these assumptions (and the corresponding algorithms) have been tested, verified and adapted several times over years before the format was finalized.

The really clever aspect to the MPEG codecs is that the specification only defines the way the replay decoder works. That intentionally leaves the door open to continual development of the encoders, so that as the understanding of the human auditory model improves, masking decisions determining which parts of the audio content are 'irrelevant information' and are therefore discarded, can be further optimised. All the encoder has to do is comply with the output data formatting expected by the receiving decoder. So different coder designers can make different decisions and prioritise different aspects in their algorithms.

And that's why certain people can hear stuff and other don't.

It's the familiar Gaussian or normal distribution model. The vast majority of people's hearing acuity sits in the middle of the curve. As you move out towards those with better or poorer hearing than 'average' their number rapidly diminishes, but there will always be some who are significantly better or worse than the 'average'. I don't know the exact percentiles for the different MP3 bit-rates but, it is designed to work for most people, most of the time... and it does that very well.

After many years of adoption, it seems that with a good encoder, a 256/320kps bitrate is definitely good enough, and 128kps holds water in many cases for most people, as theorized. No surprise most people don't complain!

I think that's a very good generalised summary.

[The Elf]

I have no problem with lossy formats as end-destination, consumable formats, but anything that converts back from a lossy format to a lossless format should be made illegal!

I know of one 'once mighty' CD recovered from MP3s that nearly made it back into mainstream production, but I won't name names...

[Hugh Robjohns]

I have no problem with lossy formats as end-destination, consumable formats, but anything that converts back from a lossy format to a lossless format should be made illegal!

Totally agree.

I spend the last few years of my BBC Training lecturer time trying to educate radio and TV production staff about why they should not use MP3s (and other lossy codecs) as source material... it was like hitting my head on a brick wall!

I kept getting the line: 'but it sounds alright on my phone and on the studio speakers, so it must be fine....'

They just couldn't or wouldn't grasp the idea that there are further lossy codecs in the transmission chain, so although it sounded okay in the studio, it becomes mangled to the real end-user at home or in the car if they're listing on DAB or DTT. I could demonstrate the effects of codec concatenation, and they could all hear the results... but they refused to join the dots with their own practices!

Check out Steve Wright's afternoon show... he's forever playing lossy-codec audio clips grabbed off the internet.

[Mike Stranks]

Check out Steve Wright's afternoon show...

I'd rather not if it's all the same to you... teeth... edge...

And on the subject of horses for courses, I sometimes have to prepare spoken-word files for hosting on sites with very limited space available. As long as the source is good and clean I find I can generate MP3s at 16KHz, 32 kbps (Lame Encoder) and have them still sound perfectly acceptable for their intended use.

[Tim Gillett]

I occasionally do the same Mike, when I need to send demos to a customer or friend via email where here there's a 10 Mb size limit. It can be a fun challenge to pull every trick before encoding to make it sound as if it's not a low bitrate mp3. I think it was Mandy Parnell who said something like "that's real mastering", or at least what mastering used to be - making a lower fidelity delivery format still sound good.

[Mike Stranks]

I occasionally do the same Mike, when I need to send demos to a customer or friend via email where here there's a 10 Mb size limit. It can be a fun challenge to pull every trick before encoding to make it sound as if it's not a low bitrate mp3. I think it was Mandy Parnell who said something like "that's real mastering", or at least what mastering used to be - making a lower fidelity delivery format still sound good.

Not meaning to be patronising Tim, but you do know there are apps available where you can send fairly chunky data sets for free? (I don't mean Dropbox, Googeldrive et al...)

I've used several, but of late I've settled on WeSendIt https://www.wesendit.com/ as it's reliable, simple to use and, most importantly, easy for the recipients to access the sent files.

Just a thought...

[Tim Gillett]

Not at all, Mike, thanks for the tip.

[CS70]

I occasionally do the same Mike, when I need to send demos to a customer or friend via email where here there's a 10 Mb size limit. It can be a fun challenge to pull every trick before encoding to make it sound as if it's not a low bitrate mp3. I think it was Mandy Parnell who said something like "that's real mastering", or at least what mastering used to be - making a lower fidelity delivery format still sound good.

Not meaning to be patronising Tim, but you do know there are apps available where you can send fairly chunky data sets for free? (I don't mean Dropbox, Googeldrive et al...)

I've used several, but of late I've settled on WeSendIt https://www.wesendit.com/ as it's reliable, simple to use and, most importantly, easy for the recipients to access the sent files.

Just a thought...

As an incidental - it's funny that there are some unexpected cases where actually that doesn't work, for surprising reasons.

A couple years ago I did some video work for a middle-eastern singer, who wanted to send the finished product back to radio services in his home country. The video came out well and very high quality - but It turned out that the main communication system in his country is some sort of 3G or even Edge mobile network, so the video size had to be incredibly small to allow the majority of people to watch it.

I spent more time working the the coloring so that I could get it to the required size without it looking like lego blocks, than actually filming and mounting it.

We're lucky in our little worlds

[Drew Stephenson]

I spent more time working the the coloring so that I could get it to the required size without it looking like lego blocks, than actually filming and mounting it.

Ugh. Tell me about it, restricted bandwidth is a major bane of my working life at the moment. You spend ages (well, a decent amount of time) getting a nice, clear, good sounding video, and then you have to compress the snot out of it to get it round the network and the colours go funny, the sound gets ugly, and everyone looks like they've been trying Trump's make-up cabinet.

[CS70]

I spent more time working the the coloring so that I could get it to the required size without it looking like lego blocks, than actually filming and mounting it.

Ugh. Tell me about it, restricted bandwidth is a major bane of my working life at the moment. You spend ages (well, a decent amount of time) getting a nice, clear, good sounding video, and then you have to compress the snot out of it to get it round the network and the colours go funny, the sound gets ugly, and everyone looks like they've been trying Trump's make-up cabinet.

Make it black and white and it'll compress in a breeze!