Help explain intersample overs, please?

[L5730]

@bennetng I'm in agreement with where you are going with this.
It's like when essentially the same 24/96 master is used to create the CD and a vinyl release. Then folks come along showing waveforms, DR stats and whatnot trying to demonstrate that the vinyl version is in some way superior.
Now, the distortions imparted by the process of cutting to vinyl and playing that medium back, the effect of pre-amp etc. all may impart pleasant qualities to the sound, but we should pretty much all recognise that the CD is likely much closer to the original master file. Ian Shepherd has a video demonstrating exactly this with something he did.

For better quality audio it's an entire process. Not ballsing up the recording (we'll assume they are at least competent in recording techniques). Mixing with headroom and at floating point export. Mastering with headroom to prevent ISPs in the final output. This should be for each track, making them the sonic best they can be.

As end consumers we only have a limited tool set. There are other threads which discuss the merits of the DR meter and stats. Data from these tools can indicate things, but we really need to get to the nitty gritty of listening to the music to hear whether or not issues arise - at the moment. Perhaps more intelligent algorithms will be developed which may be able to better analyse the quality of a track based on numerous factors. It starts getting into the territory of what is music, what is noise and what was the intended sound?

 

[bennetng]

Perhaps more intelligent algorithms will be developed which may be able to better analyse the quality of a track based on numerous factors. It starts getting into the territory of what is music, what is noise and what was the intended sound?

Like These?

From iZotope:
https://techblog.izotope.com/2017/0...lavalier-microphone-noise-with-deep-learning/

...and Celemony:

 

[L5730]

Like These?

From iZotope:
https://techblog.izotope.com/2017/0...lavalier-microphone-noise-with-deep-learning/

...and Celemony:

I guess. But perhaps something like if we were to upsample a file to a silly high sample rate, we could overlay and see the ISP or a closer approximation of the actual waveform rather than sample points (Adobe Audition does this anyway). Couple in some cleaver analysis to observe the differences between the two versions. I dunno. It gets complicated and my mind is wandering...
Trying to assign a numeric 'quality' value to audio is probably a bit of a folly. Anything with mangled and garbled synthetic or otherwise sounds which were created on purpose are just going to flag up - despite them being the actual intention of the artist. Take Pink Floyd for a loose example. Eggs frying, farm animals - oh they must have accidentally caught these odd sounds on tape when they were on location recording in an old barn. Nope, these sounds are there on purpose. An orchestral rendition of Ticket to Ride...well...

 

[bennetng]

Trying to assign a numeric 'quality' value to audio is probably a bit of a folly. Anything with mangled and garbled synthetic or otherwise sounds which were created on purpose are just going to flag up - despite them being the actual intention of the artist.

IMO finding bad quality recordings with predefined algorithms or even AI is similar to virus scanners, false positives/negatives are inevitable. Even with human listeners, different people can have different opinions and preferences. Restoration tools also have adjustable settings so it implies human judgment is required.

 

[restorer-john]

Looking at very early CDs, many had plenty of peak headroom. Highest sample peaks at -3dB or even lower were not that uncommon. Sure, some/many had pre-emphasis, but we'll leave that for another thread.

I have tons of the first release CDs and yes, they pretty much all had plenty of headroom. And many of the early releases did indeed have pre-emphasis, which as far as the listener is concerned, they are perfectly unaware of its operation. All the first machines pretty much employed de-emphasis in the analog domain and by the time digital filtering and O/S hit the market, many were doing the de-emphasis in the digital domain.

There is nothing inherently evil about pre/de-emphasis.

 

[MRC01]

...There is nothing inherently evil about pre/de-emphasis.

The most frustrating thing about pre-emphasis is when the CDs using it don't set the proper flags (something about flagging in the TOC versus the individual track), so CD players don't know it's there, so they don't apply the de-emphasis curve, and the treble will melt your ears. That's not evil, just frustrating.

It's nice to have something (whether a DAC, EQ, or software) that lets you push a button to apply it manually. It's a simple slope, should be easy to program manually into any EQ.

 

[Hayabusa]

As for looking for flat tops to verify clipping, no flat tops does not mean something is not clipped. The flat tops can be easily eliminated visually with something like a high pass filter at 10Hz, or simply ran a clipped song through a cassette deck like this:
https://www.audiosciencereview.com/...kamichi-dragon-cassette-deck.5595/post-124666

The frequency content of the error of the isp's its much higher than 10Hz, so a 10Hz low pass filter will not really help.

 

[bennetng]

The frequency content of the error of the isp's its much higher than 10Hz, so a 10Hz low pass filter will not really help.

I said "high" pass filter in the original post, and this comment is for those who look for clippings by merely looking at flat tops in waveform, which is an unreliable method.

 

[Hayabusa]

I said "high" pass filter in the original post, and this comment is for those who look for clippings by merely looking at flat tops in waveform, which is an unreliable method.

Typo: I meant high pass...

 

[restorer-john]

The most frustrating thing about pre-emphasis is when the CDs using it don't set the proper flags (something about flagging in the TOC versus the individual track), so CD players don't know it's there, so they don't apply the de-emphasis curve, and the treble will melt your ears.

Do you have any actual examples you own (CDs) of that occurring? The CD standard precluded any disc sold bearing the compact disc logo not to have the correct flag set and all CD players contained appropriate circuitry to de-emphasize said discs.

Sure, once CD-ROM based 'rippers' came along, things went AWOL fast, but an actual dedicated player not picking up a de-emph flag, I've never seen it, and I have a lot of CD players...

 

[MRC01]

Do you have any actual examples you own (CDs) of that occurring? The CD standard precluded any disc sold bearing the compact disc logo not to have the correct flag set and all CD players contained appropriate circuitry to de-emphasize said discs.
...

This list shows albums that implement the flag inconsistently: some only in TOC, others only in track, some in both. PE is rare, but one example I own is a disc called "The Digital Domain - a Demonstration" which has tracks having PE but no flag.

 

[restorer-john]

but one example I own is a disc called "The Digital Domain - a Demonstration" which has tracks having PE but no flag.

That disc has 4 tracks specifically designed to test the test emphasis error (deviation) and they are subcode flagged with and without de-emphasis to trigger the de-emph circuit. Tracks 22,23,24,25.

Most early machines simply switched in the de-emphasis time constant via the feedback loop of the buffer stage like this.

 

[Doodski]

That disc has 4 tracks specifically designed to test the test emphasis error (deviation) and they are subcode flagged with and without de-emphasis to trigger the de-emph circuit. Tracks 22,23,24,25.

View attachment 51845

Most early machines simply switched in the de-emphasis time constant via the feedback loop of the buffer stage like this.

View attachment 51847

That service manual schematic appears a lot like a Sony service manual. Do you have a repository of service manuals you collected?

 

[mansr]

That service manual schematic appears a lot like a Sony service manual.

Sony CDP-101, to be specific.

Do you have a repository of service manuals you collected?

Doesn't everybody?

 

[Doodski]

Sony CDP-101, to be specific.


Doesn't everybody?

I see the allure of having service manuals au plenty when servicing product and have had cabinets of them in the past from warranty service although why keep a large repository of discontinued product manuals that will probably never be used for service now? If I did have them I would be inclined to post them on the internet. I'm not a collector of anything really and I don't get collecting stuff like this. Is it sentimental or just the heck of having them?

 

[restorer-john]

I'm not a collector of anything really and I don't get collecting stuff like this. Is it sentimental or just the heck of having them?

I have an enormous collection of vintage HiFi and also particular interest in Compact Disc, especially the 82-92 period. There are at least 5 CDP-101s in a single large carton in my storeroom waiting for restoration and a few here, so that high res pdf manual is better (for my eyes) than my printed copy.

My NAS has hundreds of service manuals and I can put my finger on a part or schematic within a few seconds. So much better than the old days ratting through a filing cabinet looking for a manual, only to find it has been "borrowed" and never returned. Need to order a new printed $30 one and wait 6 weeks from Japan etc...

Each of these folders has hundreds of files in some cases.

I take the view that the internet is not a library for this documentation, it is a resource that could be gone tomorrow. Nothing is permanent, so get what interests you and keep it locally.

 

[Doodski]

So much better than the old days ratting through a filing cabinet looking for a manual, only to find it has been "borrowed" and never returned.

Besides a electronics study I also have a study in architectural mechanical drafting behind me and that gives me a very strong respect for any documents and service manuals. I especially love those Sony rice paper service manuals Whoever ended up with the cabinets of service manuals received detailed notes in the manuals about all service bulletins, mods, common repairs and those pesky dog repairs too. They would simply need to open the manual and it would all be laid out in super bold easy to read handwriting. Without documentation things are ineffective. Docs must be respected! ;P

 

[captain paranoia]

Intersample overs can happen because reconstructing the analog wave from the digital sampling points, is an infinite series for every sampling point.

If you're using an Infinite Impulse Response filter, maybe. If you're using a Finite Impulse Response filter, you can find the gain of the filter (taking the abs values of the FIR coefficients), and scale the coefficients to ensure the filter gain does not exceed unity. Then, regardless of the input stream, the output will never exceed FSD. The output is only dependent on the set of n samples currently passing through the n stages of the filter.

 

[restorer-john]

I especially love those Sony rice paper service manuals

The old Sony schematics are, in my opinion, the finest schematic drawings in audio (and video/tv) I have ever seen. To me, they are as beautiful as any artwork and much more interesting.

 

[Doodski]

The old Sony schematics are, in my opinion, the finest schematic drawings in audio (and video/tv) I have ever seen. To me, they are as beautiful as any artwork and much more interesting.

I agree 100%. I enjoyed using them every time and treated them like my precious ;P Some people just don't get it though. I've worked with peeps that think they are a cheaper form of paper when they are fine and silky smooth for the ink so the text and lines are the highest quality possible for our viewing pleasure