DR Dynamic Range

[Arconada]

You simly can't push a high dynamic signal through a low dynamic channel without loosing information

 

[marcb]

You simly can't push a high dynamic signal through a low dynamic channel without loosing information

You weren’t losing anything with Dolby/DBX. The problem was you couldn’t accurately recreate the original. Imagine flac with errors.

 

[Arconada]

You weren’t losing anything with Dolby/DBX. The problem was you couldn’t accurately recreate the original. Imagine flac with errors.

So what it is the difference?

 

[marcb]

So what it is the difference?

Again, Mp3 is lossy digital compression algorithm designed to throw out bits to significantly save storage space. The trade-off is the bits are lost forever.

In simple terms, DBX/Dolby utilized analog dynamic range compression/expansion to reduce signal-to-noise ratio. The problem is they just don’t (can’t) do a perfect job of ”re-expanding’ the signal - resulting in artifacts, pumping, etc. (particularly DBX).

 

[Arconada]

Again, Mp3 is lossy digital compression algorithm designed to throw out bits to significantly save storage space. The trade-off is the bits are lost forever.

In simple terms, DBX/Dolby utilized analog dynamic range compression/expansion to reduce signal-to-noise ratio. The problem is they just don’t (can’t) do a perfect job of ”re-expanding’ the signal - resulting in artifacts, pumping, etc. (particularly DBX).

You mean loosing detail is not the same as loosing information? Think about it, when you want to save a signal with a 100 dB dynamic range you will need much more storage space than when you save a 50 dB dynamic range. It simply means you will loose information when you save a signal with a 100 dB dynamic range in a "transmission channel" that only allows for 50 dB

 

[marcb]

You mean loosing detail is not the same as loosing information? Think about it, when you want to save a signal with a 100 dB dynamic range you will need much more storage space than when you save a 50 dB dynamic range. It simply means you will loose information when you save a signal with a 100 dB dynamic range in a "transmission channel" that only allows for 50 dB

That’s not at all what’s happening with DBX and Dolby noise reduction. You may be ”altering” information, but you’re not “losing” anything. If you want to call that “lossy”, you could say that about pretty much any analog recording/playback process.

The purpose of DBX wasn’t to squeeze 100db into 50db. Squeezing the dynamic range of the signal was only a means to an end. The end purpose was to lower the tape noise floor on playback from, say, -40dB to -80db and thus increase the effective dynamic range of the signal. The dynamic range of the signal itself is unchanged at the end of the processing; it’s just that the noise is much lower relative to the signal after the expansion (or de-compression). That’s why it’s called noise reduction.

What you’re talking about is more like the RIAA curve - which uses a similar “attenuation/boost” concept with a goal of creating playable grooves and optimizing the amount of music which could be cut on an LP side. But that’s where the analogy ends and, again, the RIAA pre-emphasis/de-emphasis process is not lossy either.

 

[par4ken]

Oh, but you are wrong! DBX and Dolby are lossy.

Disagree totally, neither is lossy. Only the noise is reduced as the signal level on the tape is increased during the quiet spots and the level is reduced back to the original value on playback. Assuming that the equipment is calibrated properly the process is seamless. And we are talking about Dolby noise reduction not Dolby Digital.

 

[MidiMagic]

There is a difference between when analog audio starts to clip and when digital audio start to clip. It explains why CDs are brickwalled much more than LPs.

When an analog device clips slightly, it adds a slight bit of second harmonic distortion to the recording in only the loudest notes. Often this is not even noticed when the recording is played. Some guitar players actually use it in their sound.

When a digital device clips slightly, it sounds like a train wreck. No clipping at all can be tolerated in a digital recording.

The trade-off in setting the 0dB point in a 16 bit recording is threefold:

- A higher 0dB point lowers the noise floor.
- A higher 0dB point reduces headroom before clipping.
- A higher 0dB makes the recording sound louder.

The problem is that the record companies decided on the 0dB to provide louder airplay. Then they had to do something to prevent clipping.

A brickwalling method I have used actually introduces some 2nd harmonic distortion to lower the clipped waveforms below clipping.

Part of the problem with the early CDs was that they did not correctly apply an inverse RIAA curve to recordings already mastered for records.

I have some brickwalled vinyl. The most blatant ones i can remember are most recordings by Sweet and Charlie Dore.

 

[Arconada]

I can't discuss with you if you don't understand that loosing detail is the same as loosing information. Loosing information is lossy. And if you have a transmission channel with a certain bandwith and signal to noise ratio, you can't transmit a signal with a higher SN-ratio without compressing and expanding it. Since you agree you are loosing detail in that process, you must conclude the method is lossy. If you keep holding on to the idea it's not, it is best to stop here.

 

[MidiMagic]

Are we talking about losing information (it's gone) or loosing it (letting it loose to be found)? Speling matters (sic).

 

[par4ken]

I can't discuss with you if you don't understand that loosing detail is the same as loosing information. Loosing information is lossy. And if you have a transmission channel with a certain bandwith and signal to noise ratio, you can't transmit a signal with a higher SN-ratio without compressing and expanding it. Since you agree you are loosing detail in that process, you must conclude the method is lossy. If you keep holding on to the idea it's not, it is best to stop here.

I understand what you are saying but we are getting hung up with semantics. By your definition an imperfect analogue recording would be considered lossy. A CD made from a higher rez source and down sampled would be considered lossy, I get that. By digital audio standards (of usage) the term lossy is reserved for formats like MP3 in which the algorithm actually discards bits of information that it deems to be unimportant. Down-sampled digital audio or an analogue recording would still be considered to be lossless.

 

[DaveLaPorte53]

Brickwalling simply has to do with the dynamic range of the transmission medium or the listening area. FM radio allowed for no more than 30 dB dynamic range, AM radio even less. Nowadays brickwalling is applied to music played at dance events;. lots of noise, no room for details. A brickwalled piece of music is not meant to be played at home. You simply got the wrong version

19th century?
I thought you were talking about high-end audio.

I see high end audio components and they are use tubes, what does that tell you?

 

[ssully]

I find that the DR meter is generally misleading for multichannel audio--all it does it take the DR value of each individual channel, add them together, then divide by the # of channels to generate an average. As you say, this is obviously an issue for quad albums with an empty center/sub as the average reading is being skewed by those two extra channels with a DR value of 0.

Indeed. Any format where there are , by design or common practice, large disparities between the channels, is going to confound the DR meter unless the disparity is taken into account.

The DR meter is also unreliable for vinyl, and probably just should be scrapped generally. Better methods have been devised.

see for example:
https://arxiv.org/pdf/1312.0538.pdf

 

[J. PUPSTER]

Really love these DR threads - they really kick the hornet's nest!

 

[ssully]

This is only albums with DSF I believe as the conversion does not make an empty channel.

You can get the correct SACD rip DR by by removing the silent channel then add a new one and the DR value is correct. (Music Media Helper has a tool to fix the silent channels).

For FLAC the silent channel(s) is empty and is ignored in the DR calculation.

A few of us fixed most the DR values for SACDs listed in the QQ Hires DR report recently

https://www.quadraphonicquad.com/files/QQ Hires Poll with DR.pdf

There are mixes where one, two, or three channels have been 'slammed' but the others have not. The usual DR meter reading for such a recording is misleading.

 

[Arconada]

I understand what you are saying but we are getting hung up with semantics.

I replied to your outcry that digital mastering is often CRAP, as you call it. I felt that I needed to remind you that analog music is also brickwalled when necessary, A lot of digital music passed my computer and I can confirm that most of this music is not brickwalled. If music is brickwalled it has been done for a purpose, namely noisy environments. So if you have brickwalled music where you expect it not to be, buy a proper version. The hailing of analog music over digital also is not befitting. Nowadays a vinyl disc pressing starts with a digital recording, so nothing is gained. Subsequent analog processing is lossy in the sense that you cannot recreate the original from the result, as you can't in MP3

 

[Soundfield]

How about counting the number of ones in the digital file? Clearly recordings with a predominance of lovely ones rather than those with lots of those feeble zeros would be of much higher quality.

 

[marcb]

Indeed. Any format where there are , by design or common practice, large disparities between the channels, is going to confound the DR meter unless the disparity is taken into account.

The DR meter is also unreliable for vinyl, and probably just should be scrapped generally. Better methods have been devised.

see for example:
https://arxiv.org/pdf/1312.0538.pdf

Agree with everything here except the part about DRs with vinyl. They may not be as reliable as with digital, but (if the needledrop is properly declicked) they are still fairly reliable and repeatable - more importantly can still be useful in evaluation of a particular vinyl cut vis vis other digital or vinyl masterings.

 

[marcb]

I replied to your outcry that digital mastering is often CRAP, as you call it. I felt that I needed to remind you that analog music is also brickwalled when necessary, A lot of digital music passed my computer and I can confirm that most of this music is not brickwalled. If music is brickwalled it has been done for a purpose, namely noisy environments. So if you have brickwalled music where you expect it not to be, buy a proper version. The hailing of analog music over digital also is not befitting. Nowadays a vinyl disc pressing starts with a digital recording, so nothing is gained. Subsequent analog processing is lossy in the sense that you cannot recreate the original from the result, as you can't in MP3

By your logic, EVERYTHING is lossy. Neither analog nor digital (as it currently exists) captures the original.

 

[ssully]

Agree with everything here except the part about DRs with vinyl. They may not be as reliable as with digital, but (if the needledrop is properly declicked) they are still fairly reliable and repeatable - more importantly can still be useful in evaluation of a particular vinyl cut vis vis other digital or vinyl masterings.

This made the rounds a few years ago...maybe you missed it?

https://productionadvice.co.uk/tt-meter-not-for-vinyl/