16 bit files almost unlistenable now...

[MetalNuts]

Just now, Summit said:

 

I don’t think what MetalNuts wrote it truism and I disagree on that PCM sucks. I think you need to understand that the quality of a recording matters more than the resolution or format. The quality of a recording depends on so much more than the final product and how it is packaged. That doesn’t mean that “resolution” and format are irrelevant, just less.

 

If the quality of the recording is bad no stereo or format going to make it sound good.    

 

Yes, that is what I meant.  A good recording, irrespective it being made vinyl, redbook, high rez files or DSD, it sound good.

 

However, resolution should be able to enhance the quality.  For music files, it may not be that obvious to someone, but I do not believe anyone will dispute the quality in photography and videos that higher resolution can bring.

[mansr]

7 hours ago, diecaster said:

MP3 bit depth is always 16 bits.

That's not quite true. The output of the inverse transform generally has a precision exceeding 16 bits even if the input does not. Truncating it back to 16 bits increases the distortion somewhat. The standard doesn't specify how an encoder should handle the input resolution. Best results are obtained by doing the transform in 32-bit (or better) precision, then encoding the output as closely as the format permits. The end to end performance depends on the complexity of the signal. Simple signals are coded more accurately than complex ones. That said, don't expect mp3 to exceed 16-bit precision with anything resembling music.

[Audiophile Neuroscience]

47 minutes ago, MetalNuts said:

 

Yes, that is what I meant.  A good recording, irrespective it being made vinyl, redbook, high rez files or DSD, it sound good.

 

However, resolution should be able to enhance the quality.  For music files, it may not be that obvious to someone, but I do not believe anyone will dispute the quality in photography and videos that higher resolution can bring.

 

Yes true. You can polish gold but you can't polish a turd. The more you resolve crap the crappier it becomes.

[Andyman]

5 hours ago, esldude said:

Here is a chart showing quality in one simple graphic.  No one is claiming 128 kbps is transparent vs lossless.  But they get near that once you starting going above 128 kbps.  Therefore none of the formats have a decisive advantage at the upper rates.   Opus btw is an open source codec from xiph.org.  Soundcloud can/does use Opus. 

 

http://opus-codec.org/comparison/

 

 

Hi Dennis.

Not sure about this graph. Whilst the vertical scale is somewhat arbitrary, it does seem to imply that with AAC (amongst others) there is very little difference between 32 (and a bit) and 128. I'd be surprised if that were true.

[audiventory]

29 minutes ago, Andyman said:

Here is a chart showing quality in one simple graphic.

 

To estimate quality, it is necessary to define measurable value (distortions, method of audio perception testing, etc.).

[sandyk]

44 minutes ago, Andyman said:

 

Hi Dennis.

Not sure about this graph. Whilst the vertical scale is somewhat arbitrary, it does seem to imply that with AAC (amongst others) there is very little difference between 32 (and a bit) and 128. I'd be surprised if that were true.

It may be true with AM radio ?

 

 Is this a first from Dennis ?

Dennis (of all people !) is ranking SUBJECTIVE reports of Audio Codecs  Sound Quality!

 

[fas42]

4 hours ago, GUTB said:

 

Early digital recording is mostly trash. Indeed, even modern digital recording is trash -- that's why we audiophiles are stuck in our audiophile label ghetto that use high end digital recording techniques and employ talented studio engineers. 

 

Native DSD recordings are the best digital has to offer. Unfortunately digital doesn't like 1-bit sample rate because you can't do logic operations on 1 bits (without padding, ie, DSD-Wide), so these native DSD recordings can only do small ensemble / solo type music.

 

PCM sucks and nothing is going to fix it. However, it can be pretty good starting at 176/192. Redbook sucks.

 

As people are pointing out, you're completely wrong. Of course, going to a typical high end audio show would make one believe that sort of thing - it takes exposure to a cheap rig getting the sound right to appreciate what's possible; the vast universe of all the recordings done to date is then available for one's listening pleasure.

[Andyman]

42 minutes ago, audiventory said:

 

To estimate quality, it is necessary to define measurable value (distortions, method of audio perception testing, etc.).

 

Whoah there. I didn't post the graph but was questioning it (bloody Dennis causing me trouble!)

[audiventory]

7 minutes ago, Andyman said:

I didn't post the graph but was questioning it

 

Looks like I done something wrong when quote.

[Audiophile Neuroscience]

1 hour ago, Andyman said:

the vertical scale is somewhat arbitrary

 

Very

[PeterSt]

 

4 hours ago, sandyk said:

Some of the earlier recordings were very good indeed, but it took much more recent DACs and output stages with much improved S/N to do them justice,

 

I have to agree with this.

 

4 hours ago, sandyk said:

as they were recorded at maximum  levels well below those used these days,

 

Like with Roberta Flack eh ? I mean, I see the mistake made regularly in this thread - the recordings were from long before 1983 (like 1973 (or maybe actually 1969) for this Roberta Flack).

 

4 hours ago, sandyk said:

as they were recorded at maximum  levels well below those used these days, and had little (if any) compression.

 

So transferred to digital, yes. And the key is to find the transfers which were just new as close as possible to 1983. I mean, back in those days nobody had learned to destroy recordings (by means of too much compression).

 

4 hours ago, sandyk said:

" Roberta Flack-Killing Me Softly"  was a good example of the earlier CDs that earlier players were unable to do justice to.

 

So, maybe. Fact (for me, over here) is that the oldies by now sound the best. Not "a few" but most. And "the best" means better than any later remaster or new recording. Say it is the sad truth.

 

Something else is that we never know how something sounds, unless we can compare. Example :

The right hand one you may not know, but it comes from this one :

 

Compare Track 01 of the left hand grid with Track 04 of the right hand grid.

The former should be the "original" while the latter is a collection but with a better transfer hence even higher DR (this is the |12.5| figure). Indeed it is so that the former has a relative very low "maximum level" (only uses just about half of the digital headroom) while the latter is better on it. So in this case the collection album contains the better transfer (of that particular track). Btw, I always play this one and it sounds plainly superb.

 

As a bonus, this one :

 

 

 

As you see, it can always be done better. So the guy who did this squeezed out 19.2 (Track 06). But all he did was taking into account the available headroom (with some "interpretation" we can see he did not do this per track).

So should this be the best version ? maybe. Is that because it is 24/96 ? of course not. It has the largest DR though, so it should be good.

But it is vinyl, so maybe not ... or it is - depending on your general preference.

 

Peter

 

 

 

 

 

 

[gmgraves]

11 hours ago, GUTB said:

 

This is just a truism that hides the reality that high resolution is better.

 

Early digital recording is mostly trash. Indeed, even modern digital recording is trash -- that's why we audiophiles are stuck in our audiophile label ghetto that use high end digital recording techniques and employ talented studio engineers. 

 

Native DSD recordings are the best digital has to offer. Unfortunately digital doesn't like 1-bit sample rate because you can't do logic operations on 1 bits (without padding, ie, DSD-Wide), so these native DSD recordings can only do small ensemble / solo type music.

 

PCM sucks and nothing is going to fix it. However, it can be pretty good starting at 176/192. Redbook sucks.

You are aware that all DSD recordings have to be converted to LPCM to edit them, after which they are converted back to DSD?

[firedog]

1 hour ago, gmgraves said:

You are aware that all DSD recordings have to be converted to LPCM to edit them, after which they are converted back to DSD?

Not necessarily. They can be converted to tape, edited, and converted back to DSD. Or the pyramix system can be used, which only coverts the very small place of the edit itself, and leaves the rest of the recording alone. 

Both of the above methods are in use. Obviously they aren’t used for multitrack recordings and others that require extensive processing.

[GUTB]

2 hours ago, gmgraves said:

You are aware that all DSD recordings have to be converted to LPCM to edit them, after which they are converted back to DSD?

 

Hence native DSD recordings. You can only do basic mixing/editing on these without destroying the benefit of DSD.

[sandyk]

9 hours ago, PeterSt said:

So, maybe. Fact (for me, over here) is that the oldies by now sound the best. Not "a few" but most. And "the best" means better than any later remaster or new recording. Say it is the sad truth.

 Peter

 I have  a copy of  the master of " Peggy Lee-Fever" before they mixed in all the other stuff and it sounds great.

 The eventual CD sounds pathetic in comparison. Perhaps the S/N of the mixers of the time, or too many channels in use were the reason for much of the degradation ?

 

Alex

[fas42]

1 hour ago, sandyk said:

 Peter

 I have  a copy of  the master of " Peggy Lee-Fever" before they mixed in all the other stuff and it sounds great.

 The eventual CD sounds pathetic in comparison. Perhaps the S/N of the mixers of the time, or too many channels in use were the reason for much of the degradation ?

 

Alex

 

Alex, are you talking about the 'standard' version of that song we all used to heard on the radio, over the years?

[gmgraves]

4 hours ago, firedog said:

Not necessarily. They can be converted to tape, edited, and converted back to DSD. Or the pyramix system can be used, which only coverts the very small place of the edit itself, and leaves the rest of the recording alone. 

Both of the above methods are in use. Obviously they aren’t used for multitrack recordings and others that require extensive processing.

Pedanticism, thy name is Firedog! The point was that DSD cannot be edited as DSD, It has to be edited in another format. There is nothing that says that the DSD file couldn't be converted to 33.3 RPM stereo lacquer transcriptions to be A/B roll edited (like video tape) to a third 33.3 RPM stereo lacquer transcription, before being converted back to DSD. But I'm reasonably sure that a similar number of record companies do that as transfer their DSD recordings to analog tape to edit them. 

[gmgraves]

3 hours ago, GUTB said:

 

Hence native DSD recordings. You can only do basic mixing/editing on these without destroying the benefit of DSD.

Believe me, I'm well acquainted aware with that particular shortcoming of the format. 

[gmgraves]

23 hours ago, fas42 said:

I don't have a problem with mp3 at all - at one stage I got very interested in trialing this format, and played quite a bit with the settings of LAME, to see what happened with the SQ. Yes, one can hear the alteration, but it's not a disaster - perfectly acceptable, in the same way as a slightly different mastering is acceptable. Creating a diff file showed how little was lost, compared to the large variations that can occur for other reasons.

 

If I was forced to listen to only good quality mp3 from now on, it wouldn't be hell on earth ...

Good luck with that, Frank! I'm glad somebody doesn't have a problem with it, but that somebody isn't me! I find anything lower than 128 kbps to be unlistenable! The combination of pre-echo and ringing coupled with the noise modulation makes for a very unsatisfactory listening experience. at 192 kbps, I find listening on speakers to be acceptable but it requires a bit rate of 320 kbps to be acceptable to me on headphones. BBC3 is using AAC nowadays to stream classical music and I have to say, I have heard no artifacts on either speakers or headphones yet.  

[fas42]

33 minutes ago, gmgraves said:

Good luck with that, Frank! I'm glad somebody doesn't have a problem with it, but that somebody isn't me! I find anything lower than 128 kbps to be unlistenable! The combination of pre-echo and ringing coupled with the noise modulation makes for a very unsatisfactory listening experience. at 192 kbps, I find listening on speakers to be acceptable but it requires a bit rate of 320 kbps to be acceptable to me on headphones. BBC3 is using AAC nowadays to stream classical music and I have to say, I have heard no artifacts on either speakers or headphones yet.  

 

Yes, I was talking of using the LAME encoder at the very highest quality settings, and 320 kbps, when doing these experiments - it just demonstrates that the process can get the job done, if you absolutely have to have compressed sound.

 

Also, a key step is that the mp3 is decompressed offline - I'm actually playing a WAV file, which has been through the encoding/decoding mill - some people will say this makes no difference, but I found otherwise, on the gear I was using.

[gmgraves]

13 minutes ago, fas42 said:

 

Yes, I was talking of using the LAME encoder at the very highest quality settings, and 320 kbps, when doing these experiments - it just demonstrates that the process can get the job done, if you absolutely have to have compressed sound.

 

Also, a key step is that the mp3 is decompressed offline - I'm actually playing a WAV file, which has been through the encoding/decoding mill - some people will say this makes no difference, but I found otherwise, on the gear I was using.

It still seems that AAC is better. The BBC3 stream is very nice to listen to on speakers and headphones!

[audiventory]

7 hours ago, GUTB said:

You can only do basic mixing/editing on these without destroying the benefit of DSD.

 

By available publications, editing 1 bit DSD to 1 bit DSD without re-modulation here is cut/merge only.

 

 

Mixing 1 bit DSD to 1 bit DSD is impossible without re-modulation.

 

Without filtering, part of dynamic range is lost. Filtering give ringing.

 

However, mixing (both PCM and DSD) can use digital filters and have ringing and amplitude/phase distortions. It may become matter comparing with filtering DSD.

 

Analog mixing have non-linear and frequency distortions, noise.

 

When record is played back PCM converted to DSD with ringing during upsampling.

 

Pure PCM R2R DAC have more issues, than DSD modulated.

 

Resume:

If record is edited, I don't see problem in accurate conversion DSD to PCM for editing comparing other issues in given context.

 

 

 

 

[esldude]

4 hours ago, gmgraves said:

Pedanticism, thy name is Firedog! The point was that DSD cannot be edited as DSD, It has to be edited in another format. There is nothing that says that the DSD file couldn't be converted to 33.3 RPM stereo lacquer transcriptions to be A/B roll edited (like video tape) to a third 33.3 RPM stereo lacquer transcription, before being converted back to DSD. But I'm reasonably sure that a similar number of record companies do that as transfer their DSD recordings to analog tape to edit them. 

Actually there are at least a few who do have a mix and tape step in the middle with DSD on each side. 

[firedog]

5 hours ago, gmgraves said:

Pedanticism, thy name is Firedog! The point was that DSD cannot be edited as DSD, It has to be edited in another format. There is nothing that says that the DSD file couldn't be converted to 33.3 RPM stereo lacquer transcriptions to be A/B roll edited (like video tape) to a third 33.3 RPM stereo lacquer transcription, before being converted back to DSD. But I'm reasonably sure that a similar number of record companies do that as transfer their DSD recordings to analog tape to edit them. 

Well, if there are multiple record labels doing exactly what I described on a regular basis, I don't think I'm being pedantic. Maybe you are unaware of them. Maybe you don't know the definition of the word "pedantic".

 

In fact it might fit better to your previous post, where you pedantically pointed out something and stated it as an untrue absolute about ALL DSD recording. 

[sandyk]

8 hours ago, fas42 said:

 

Alex, are you talking about the 'standard' version of that song we all used to heard on the radio, over the years?

Frank

 I have a copy from the master where the recording engineer states the ID number of the recording, then does a countdown of 1,2,3,4 etc.

Several members, including Kumakuma have heard this track.

 

Alex