Bandcamp offer 24 bit download - any idea why?

[JeremyFife]

Hi, I wonder if anyone here can offer some insight

I noticed that Bandcamp (music streaming, selling and download site with a good track record of paying artists a decent amount) was offering music by Adrienne Lenker in 'regular' 16/44.1 and also in a more expensive 24-bit format. I've no illusion that high-res should sound better in any way but I was intrigued.

I think Adrienne is fantastic anyway so I was happy to explore https://adriannelenker.bandcamp.com/music and I bought a song in each of the two versions. I'm pretty sure they are identical. I can't share the actual music, obviously for copyright reasons, but I did run them through MusicScope and MasVis (reports attached)
I also opened them in Audacity, inverted one and mixed them together ... resulting in a flat line (not surprised).
They sound the same to me ... brilliant, but the same.

The reports say different things, and I'm aware that I don't really understand what they are saying.

Can anyone confirm that these are basically identical, or am I missing anything? Does the 24-bit version contain any extra information? What would be the point of putting both of these versions out, apart from wanting to part the gullible from their money?

 

[DVDdoug]

Sometimes the question is, "Why not 24-bits?'

And if some people are willing to pay more for it, why not offer it?

The reports say different things, and I'm aware that I don't really understand what they are saying.

That's not unusual in the audio world!

 

[Emlin]

Hi, I wonder if anyone here can offer some insight

I noticed that Bandcamp (music streaming, selling and download site with a good track record of paying artists a decent amount) was offering music by Adrienne Lenker in 'regular' 16/44.1 and also in a more expensive 24-bit format. I've no illusion that high-res should sound better in any way but I was intrigued.

I think Adrienne is fantastic anyway so I was happy to explore https://adriannelenker.bandcamp.com/music and I bought a song in each of the two versions. I'm pretty sure they are identical. I can't share the actual music, obviously for copyright reasons, but I did run them through MusicScope and MasVis (reports attached)
I also opened them in Audacity, inverted one and mixed them together ... resulting in a flat line (not surprised).
They sound the same to me ... brilliant, but the same.

The reports say different things, and I'm aware that I don't really understand what they are saying.

Can anyone confirm that these are basically identical, or am I missing anything? Does the 24-bit version contain any extra information? What would be the point of putting both of these versions out, apart from wanting to part the gullible from their money?

Proof that hi-res is a flatlined technology.

 

[rolfrander]

If you want to support your favorite artist with a bit more money, you can. And in return you get some apparent extra value. After all, 24 bits does give you 48 dB more dynamic range .

 

[Emlin]

If you want to support your favorite artist with a bit more money, you can. And in return you get some apparent extra value. After all, 24 bits does give you 48 dB more dynamic range .

Only if it was recorded with 48dB more dynamic range, which it wasn't.

 

[kemmler3D]

Can anyone confirm that these are basically identical, or am I missing anything?

It does look like the Masvis application didn't work properly on the 24-bit track, but the music scope stuff looks identical as can be. If they are coming from the same master, the only thing that can be different on a 24-bit track is lower digital noise floor, which is almost always inaudible with 16-bit anyway.

What would be the point of putting both of these versions out, apart from wanting to part the gullible from their money?

Thinking generously, if they wanted people to be able to sample / rework / remix the track, offering 24-bits is helpful. When you edit and heavily process audio, you can pull the noise floor up to an arbitrary extent, so starting with 24-bit is a very good idea if you intend to do that.

 

[JeremyFife]

I like the 'generosity' angle, thanks all

 

[Robocop]

Most new recordings are recorded in 24 bits in studios and down sampled to 16 for sale. Why not sell them at 24. I would always buy 24 when offered. Its widely used professionally.

 

[kemmler3D]

Most new recordings are recorded in 24 bits in studios and down sampled to 16 for sale. Why not sell them at 24. I would always buy 24 when offered. Its widely used professionally.

I think they were charging extra for it. Not usually going to be able to hear the difference on playback.

 

[amirm]

but the music scope stuff looks identical as can be.

Are we looking at the same thing???

How do these look the same??? The energy spectrum looks completely different.

In general, 24-bit content indicates a master that has not yet been converted to 16 bit for streaming to masses. Since there is no way to know if the person performing that conversion can do it in a perceptually lossless way (noise shaped dither), then I prefer to get the 24 bit version. Stuff on Bandcamp is very cheap anyway so I don't mind paying extra.

 

[norman bates]

if I remember, 24 bits, it only has 2 more data bits of the audio signal than 16 bit.

There is redbook 16 bit/44k, then 24 bit / 96k, so a squeek more audio resolution and double the words.

Is this right ?

 

[voodooless]

How do these look the same??? The energy spectrum looks completely different.

I think that’s just a display issue. Notice that the scales are totally different, showing 96 dB for 16 bit and 144 dB for 24 bit content. If you scale them to match they would probably perfectly overlap.

 

[amirm]

I think that’s just a display issue. Notice that the scales are totally different, showing 96 dB for 16 bit and 144 dB for 24 bit content. If you scale them to match they would probably perfectly overlap.

Ah, you may be right. Unfortunately there is no commercial program that can analyze the effective bit depth of a music file. You need to perform a statistical analysis to detect audio signal below the noise floor.

 

[norman bates]

found this

"Data on a CD is stored as an 8 bit word. The first 7 are level steps and the 8th is ± polarity, thus you get 14 bits (eight to fourteen modulation).
A 16 bit DAC allows for oversampling and digital filtering, the goal here is to avoid brickwall filters at 20Khz.
24 bit DSP chips use 8 bits for housekeeping (math processing) and 16 bits for the signal."

 

[Mnyb]

Fun thing about bandcamp is that you sometimes get a 24bit version unknowingly i looked at the file size and then had a closer look at some uploads !

So some artist does not make any fuzz at all about this they just upload thier 24/48 original and do not bother to make any other version , why should they

 

[voodooless]

found this

"Data on a CD is stored as an 8 bit word. The first 7 are level steps and the 8th is ± polarity, thus you get 14 bits (eight to fourteen modulation).
A 16 bit DAC allows for oversampling and digital filtering, the goal here is to avoid brickwall filters at 20Khz.
24 bit DSP chips use 8 bits for housekeeping (math processing) and 16 bits for the signal."

That sounds like total nonsense. While eight to fourteen encoding is used in AudioCDs (and other media), it is done to make it easier for the laser mechanism to work reliably. In top of this we have the Reed Solomon error correction code before we actually end up with the raw PCM audio data.

Oversampling has nothing to do with the bit depth, and the DSP thing is just nonsense.

 

[solderdude]

found this

"Data on a CD is stored as an 8 bit word. The first 7 are level steps and the 8th is ± polarity, thus you get 14 bits (eight to fourteen modulation).
A 16 bit DAC allows for oversampling and digital filtering, the goal here is to avoid brickwall filters at 20Khz.
24 bit DSP chips use 8 bits for housekeeping (math processing) and 16 bits for the signal."

Data on a CD is stored using eight to fourteen modulation (EFM).
Every 8 bits (= 1 byte) is changed (using the EFM method to pits and lands (well everything around a pit is basically 'land') varying in length between 3 periods and 11 periods so a pit is minimum 3 bits long (3T) and max 11 bits (11T) long thus 8 bits(= 1 byte) and the 3+11 makes the 14 bits hence: 8 to 14.
These transition between pits and lands merely defines a flipping point between a number of consecutive, same value, bits.
The 3T pulse (the shortest length) is used to govern the disc speed and the derived clock frequency.

However, these there is also cross-interleaved Reed–Solomon code (CIRC) used (spreading data words over time and adding redundancy/parity for read error correction).
So the 8 bits (1 byte) are not all data bits are not all 'music bits'. There is error correction, track info, sync info (defines where the data starts) and other data in the 'data-stream' too.

So all 16 bits of each sample + additional data is stored a bit scattered over the surface area using more than 16 bits.

https://www.electroyou.it/image.php?id=7582

And the answer to the question of 24 bit vs 16 bits is ... people ask for it (because they know 24 bit 'sounds better') and more money/downloads can be had selling other formats of basically the same data as well.

Its a bit like selling the same book but one as a paper-back and the other one nicely bound with a beautiful cover on 'nicer' paper and a bit bigger in size.
Same content, different experience.

Note: For MQA CD indeed one loses some bit depth which is then used to expand the frequency response beyond CD quality when MQA decoded. On a regular CD you increase the noise floor a little... in reality one could better see this as 'the noise floor of the recording in described a little less accurate' as only quantization noise increased a bit.

 

[norman bates]

Ty

 

[antcollinet]

if I remember, 24 bits, it only has 2 more data bits of the audio signal than 16 bit.

There is redbook 16 bit/44k, then 24 bit / 96k, so a squeek more audio resolution and double the words.

Is this right ?

Are you comparing 24bit floating point with 16 bit fixed point, where 7 bits are used for the exponent?

24 bit fixed point as generally used in audio has 24 bits of data just as 16 bits has 16 bits of data

 

[voodooless]

Are you comparing 24bit floating point with 16 bit fixed point, where 7 bits are used for the exponent?

24 bit fixed point as generally used in audio has 24 bits of data just as 16 bits has 16 bits of data

24 bit float also has 24 bits of data it’s just that the numerical range that you can make with it is much larger, with a loss in precision as a trade off.