May-be you do not listen music at real levels or in a too noisy environnement ?
The hiss of my (ex) Ampex ATR100 with Dolby A was clearly audible. And, yes we could hear signals recorded lot under this floor.
In any case, he has produced beautiful systems to ... deteriorate the sounds and aggravate the slightest irregularity of the response curves of the tape recorders. Kinda chemotherapy.
And keeping these Racks of Dolby more or less tuned in the studio was a nightmare. One of the reason I was so happy with digital was to get rid of this junk.
Stupid ? Certainly not. One only has to see how he managed to take the cinema in hostage with a totally dishonest system , technically, when no one needed his services, thanks to digital.
Last edited:
It seems to me that if you can hear noise in the signal (lets say up to noise at 15...20dB below the signal), then you should be able to hear signal below the noise to about the same level differential. That is with analog sources, when the signal is digitized then the noise gets truncated.
A lengthy story on the Wall Of Sound where it is said JC spread the Holy Book Of Acoustical Engineering by Harry Ferdinan Olson to everyone involved.
Once upon a time ~1990, yes we did prove it. I have none of the materials used today, that ITA demonstration is a matter of historical record.
You understand the concept: a single S/N number is the sum of energy in the passband. The actual energy in the critical masking band near a single tone is far lower. The more mathematically inclined people here can likely give accurate numbers to this.
If you are interested you should be able to replicate the experiment using your own ears or a spectrum analyzer, white noise and a signal generator.
Cheers!
Howie
Hearing irregular sounds in nature background noise as they are produced by different sources than the noise is VERY different than hearing signals under the level of the constituents of that sound . With magnetic tape we have the Bias oscillator doing a great thing: lowering the tape noise to the absolute minimum possible by making the vector of the magnetic particles as random as possible. We always hear the sum of the individual vectors of the magnetic material on tape, not the sum of the particles scalar as sound will be done only by the modification of that vector sum, while its scalar is the same with the tape noise scalar.
Mathematically it's impossible to show that the instantaneous value of the sound recorded on tape is smaller than the tape hiss made by the same magnetic powder making the sound.
It's the time scale than can change things, while some signals might be too short to be perceived accurately by the brain and they'd be taken as very faint signals while they actually have larger values than the tape hiss , just too short .
Mathematically it's impossible to show that the instantaneous value of the sound recorded on tape is smaller than the tape hiss made by the same magnetic powder making the sound.
It's the time scale than can change things, while some signals might be too short to be perceived accurately by the brain and they'd be taken as very faint signals while they actually have larger values than the tape hiss , just too short .
Last edited:
I totally agree and have stated so before. But taking something with " a grain of salt" means imo something different than dismissal.
Could be a misunderstanding as this critique is not aimed at Holman but more at Shure.
As said before i trust in the correctness of the data they´ve published but wonder about the lack of information about the data sampling process.
Just as an example; Shure´s Kogen wrote about the trackability again in Audio in 1973 (the velocity graph is again included with some additional data points included) and emphasized the impact of warps on the trackability, mentioned that they did a study to find out about the warp present on recordings. This study was published in the JAES and a graph from it included in the article, mentioning 67 records and Kogen wrote about it:
"We should also state that the 67 records randomly chosen for this study were pressed within the past few years and include samples from most of the large record companies throughout the world. They should, therefore, be representative of the typical audiophile's record collection."
(J. Kogen, B. Jakobs, F. Karlov; Trackability --1973; Audio, August 1973, 18)
I would not expect to see a list of these records but it is actually more information as in the case of the for years ongoing Shure study on record velocity.
Allegedly Shure presented their study in 1973 but despite that in Anderson et al´s article in Audio from 1978 they used the velocity data from 1966/1967 and the article on warps from 1976 is referenced but not the Shure study on record velocities from 1973 (neither any other version of this work from earlier years); referenced are only the trackability articles from 1966/1967.
So it remains kind of a mystery; a possible hypothesis could be that only a few dozens of records (or even less) really had these high velocities at higher frequencies, maybe some direct-to-disc records purposely while others because of some misdoing during the cutting process.
I hope it explains why there are some concerns wrt these data points, but nevertheless if one tries to design for worst case conditions it should be incorporated.
Last edited:
That is not true , because:
So the signal is NEVER under the noise level which is just spread both in the audio band AND timescale (linked to tape speed and magnetic hysteresis and also to physiological perception laws)
Sorry, but, what if you try to not be so ... definitive ?
Our brain works like an adaptative pass band filter.
Once you focus on particular bands of frequencies, the amount of energy occupied by the background noise (hiss etc.) in those reduced bands is far below the total amount of energy of this random noise in the totality of the audible range (that we measure in signal/noise ratio numbers).
We can mimic this behavior with active pass-band filters controlling noise gates, when we want to to bring out information from a background noise that makes it unintelligible at first. There are plenty of (fake) examples in TV dramas featuring American police agencies ;-)
(Sorry for my poor English)
Please, notice that hhoyt made exactly the same answer than me. No surprise if we both had the same king of experience in real life on this subject.
Last edited:
Perhaps the sticking point here is the misleading aspect of an aggregate, single s/n number? If you take a s/n reading of the actual energy which could potentially mask a single tone, it would be far less than that single s/n figure quoted for any media. From that perspective, I would agree the signal would have to be above the masking energy in dB in order to be heard. This is the core idea behind our demonstrations, and why a 14-bit system with 84 dB of dynamic range did not have more dynamic range than a cassette. (of course there is always dither, but that is another subject)
This is the same problem I have with a single number distortion spec: the nature of which order harmonics are present have more to do with how obnoxious the distortion will be than a single number indicates.
This is data compression as applied to specs: just as with audio compression, you have to discard information to save space.
Cheers,
Howie
ps: regarding your bias comment: it does not randomize domains, it actually aligns them with the polarity present somewhere on the record head pole piece close to the trailing edge of the gap. We refer to this point in space as the edge of the bias bubble and the exact polarity recorded at that point in space depends on the instantaneous sum of audio and bias. Since HF signal acts as additional bias, the bias bubble size is dynamically modulated in size by the signal, so the actual point of record on the tape is modulated back and forth longitudinally on the tape which causes time smear to the recording. When correctly set for tape characteristics Dolby HX fixes this problem and makes a nice improvement to the clarity of recordings. Unfortunately as is the case with their noise reduction, proper operation is predicated by having the action exactly matched to the tape formulation which made consumer use problematic. In a factory replication scenario both worked very well since tape formulations were known and fixed, and record conditions could be exactly set for each recorder.
Yes, but...So Dolby just enhanced the natural sliding band process in our brains for the days when we don't like struggling to listen to music
- The transparency of the dolby units were far to be optimal (VCAs are poor performers).
- The way they used (like DBX as well) to increase the available dynamic (compression during recording/expansion during play back) just increase the inevitable non-linearities of the tape recorders by the same amount.
- Not to forget the time behavior accuracy between the coding/decoding process.
To resume, Dolby offered-us a way to reduce the perceived tape hisses during silence or low level signals, at the price of loosing the quality of the interesting musical content. We had to chose our poison: Dolby, direct engraving, 32 IPS. Each one with no free lunch.
Last edited:
Can it be proved that man ever visited to the moon to people firmly convinced it is a hoax?
T, you said it very well! Dolby was a 'band-aid' not a free lunch. Even the Dolby company knew how much 'damage' it did to the original signal. They told me this in 1970, when I was applying for a job there. My associates and I tried over the decades to get Dolby to improve their products as much as possible with some success. We found that SR was difficult to detect on typical rock music, BUT still problematic on classical. Of course, Ray Dolby thought it was 'perfect', but it wasn't. He also used Holman's stuff, because it was as good as needed to be, etc, in his own listen room. He might have spent millions on the 'acoustics' but relatively little on the electronics. He was just deaf to it.
Some info: Jam thinks my 2nd modded dac sounds better than DAC-3, and that I should package it up and sell it. Too much work to commercialize it, I'm retired.
In terms of new information, chipping away at problems often proceeds slowly. When AK4499 becomes available, we will see what can be done with that. Only way to get any more out of ESS than we already do might require a lot of DSP. In that case we would probably only be using its modulator, and if so, why not some other modulator?
I just finished doing my EMI/RFI testing on some microphones.
The microphone previously used in the application I am aiming at when placed in a 2 volt P-P field showed all sorts of noise induced from a frequency of 30 MHz. It was even worse at 100 MHz. It is actually enclosed in an aluminum case although the actual electret condenser capsule is not.
So not surprisingly extra high frequency energy does affect the audio band performance.
My microphone showed no issues whatsoever from any field I could generate! It does use a capsule with built in bypass capacitors and the 100 ohm resistor to the input of the bipolar opamp stage. For some strange reason this reduced my stress level as the alpha testing stage seems to have been successful. The Beta stage will involve 600 units in the field and I don't want to think of the costs if an issue shows up!
(At first it didn't work until I realized a diode was in backwards...)
The microphone previously used in the application I am aiming at when placed in a 2 volt P-P field showed all sorts of noise induced from a frequency of 30 MHz. It was even worse at 100 MHz. It is actually enclosed in an aluminum case although the actual electret condenser capsule is not.
So not surprisingly extra high frequency energy does affect the audio band performance.
My microphone showed no issues whatsoever from any field I could generate! It does use a capsule with built in bypass capacitors and the 100 ohm resistor to the input of the bipolar opamp stage. For some strange reason this reduced my stress level as the alpha testing stage seems to have been successful. The Beta stage will involve 600 units in the field and I don't want to think of the costs if an issue shows up!
(At first it didn't work until I realized a diode was in backwards...)
Noise can make a signal that by itself is below the threshold of detection become detectable. Noise is your friend when it comes to detecting low level signals. This might actually be part of the explanation why many prefer vinyl in spite of its many shortcomings. A reverb trail may be detectable with vinyl because of the noise, whereas with digital it remains under the threshold.
Hoyt's point is the other half of the story. Masking occurs when a strong signal swamps the surrounding hair cells. Noise does not contain a strong signal in one specific band, it is spread spectrum.
This may be counter intuitive. Please Google & ponder before throwing red herrings back.
You mean sonically? It sounds a little different vs DAC-3, less forward in the midrange (to the extent of maybe being somewhat understated there, can't easily help that with this design). Level of detail and distortion sounds virtually identical to DAC-3, IMHO (likely limited by what ESS can do). Mostly, Jam and others say mine is more listenable/enjoyable/preferred. One can enjoy digital music for longer before one tires of it, if you know what I mean. The dac still isn't up what a few of the way more expensive ones can do.
Last edited:
Where? It was about real issues to "to hear any of that". Can you quote it?Here is an example.
- Status
- Not open for further replies.