Good point maybe.
I had the three versions lined up in Cool Edit multitrack and soloed them.
Next time around I should record three versions to CD and try that.
Dan.
Not quite so silly as it may sound.
If all versions are recorded to one CD (wav format) any on the fly decoder replay issues are removed.
Dan.
Guys, you can't play mp3 >data< as music. It always is converted back to linear PCM data before the DAC can use it.
EDIT - We cross posted. I don't believe these issues exist. The decoder was designed to use as little compute resource as possible
Cooledit does not work over the data and produce a better answer because it does not need to be real time.
Last edited:
BBC will only give the url for the 320kb stream to manufacturers or aggregators. So iplayer on windows works, sonus works and some others who BBC decided were worth talking to. For legacy equipment that doesn't support the new codecs there is a 128kb feed freely available.
So unless someone finds the url if you listen with foobar or vlc (or in my case gmpc) you are blocked from HD content.
Another cross post.
I think it is still worthwhile to spend the price of a recordable cd to differentiate any fine differences, more so between MP3 decoders.
I would use CE to convert Wav to FH MP3 and back to wav, and the likes of Foobar to Lame convert wav to MP3 and back to wav.
Then burn the original wav and the two reconverted wavs to CD and take a closer listen.
Dan.
Fair enough, testing different codecs is a very valid exercise, they are not all the same. In the case of lossless I think the file size is the only difference.
As I mentioned before IIRC IBM has some patents on a lossless arithmetic compression scheme that could never be used real time, or so I was told.
There was also a famous scam a long time ago where someone claimed to have an image compression algorithm that would compress any image into a 4 or 8k file, don't remember the details.
This is the usual suggested test that has nothing to do with the context of listening to music. How about seeing how much masking a 10Hz tone does to a 5kHz tone?
The question with music generating nothing above 50Hz what makes the 5000Hz tone? I know the 100th harmonic distortion.
.
Really? What kind of SYSTEM do you listen to? Pavel has shown transducer distortion measurements and wondered why when they are so high we can still tell the difference between amplifiers. (Just for the slow it is amplifier distortion components times the distortion from the rest of the system components that yield the final distortion.)
(Now I expect Scott knows when you have just two tones they also generate two more, the sum and the difference. Now when you have say up to seventh harmonic distortion times even only fifth you have a slew of sidetones, some of which reinforce and some actually may subtract.)
Ian H. in Linear Audio 4 maintains that we can perceive .01% 19th harmonic distortion. (-80 dB.) (The guest editorial in that issue might give SY a pause.)
So if you prefer try 75 hertz and 1425 hertz. Then we can discuss the spectra of music and how that affects distortion perception. (Yes in typical recordings this will put us at -110dB and I suspect with some audiophile boom & ting ones we may approach -140 dB.)
Then we can look at Dick's comment on how very small amounts of jitter may be analysed as reducing S/S+N.
Or you can snidely dismiss this all out of hand, much easier on the analysis.
Ed, these random back and forth discussions are pointless. Do you ever wonder why two folks from either side still have never sat down TOGETHER to test and report some of these extreme claims? I would gladly sit down with you and let you try and show me the "dramatic" sonic difference between two brands of 1% resistors. I promise to be nice but you have to play fair, no peeking.
EDIT - I happen to have LA 4 right next to me. Detecting the presence of a 1kHz tone in conjunction with a 50Hz tone with NO other sound present is not relevant to listening to music. If you take a barometric microphone in a typical office building there is a huge SPL of low frequency stuff going on all the time.
Last edited:
Ed, please do not put your explanation into my mouth. I am saying something different:
1) Every well engineered amplifier has all spectral components of harmonic and intermodulation distortion far below spectral components of the speaker. No exception.
2) It is a myth that speakers have only 2nd and 3rd harmonic distortion components.
3) Harmonic or intermodulation distortion is NOT the reason why 2 well engineered amplifiers soun different. Blind track.
4) It is useless to fight with distortion numbers to 0.0000000000X% for the reason that inherent speaker nonlinearity does not allow to supply the speaker with "pure" current and/or voltage.
5) speaker-amplifier interface may result in better result with "worse" amplifier, at acoustical side.
I have been curious about the apparent differences between specmanship and what we hear... that wide gap... if there is one --
The HEADLNE specs are given in order of importance (or perceived importance). here is what I mean by Headlines :
View attachment headlines spec.pdf
Notice the Dynamic range is listed first in importance. And the number is -118dB. Lets look a little closer to that number.
View attachment Detailed spec.pdf
We see it is: 1. the weighted number. 2. It is based on mfr definition of 0 FS which is 3 v rms in this case. It is basically the thermal noise floor to max output (near clipping).
But that is not the Standard value. In fact the standard does include THD and Noise and IS the actual dynamic range to compare according to the IASA. In this case, it is 'typically' -90 dB re 0FS as shown in specs. See Standard here:
View attachment ADC Dynamic range standard.pdf
And, if we use a output level more in line with home audio -- 1 volt rms we can subtract up to 10db from these numbers. That brings the listener to 'typical' - 80dB dynamic range. And, no minimum spec is given to which up to 10db would also be subtracted for home apps.
To these realistic numbers and normalized to the standard.... It seems to me very likely it is above the audible threshold.
BUT, wait. There is more. I am comparing the ADC standard to a DAC spec.... a standard I would like to use because with Hi Res download from master which used ADC we could get exact copy.
But the Standard is also realistic on that DAC side as requiring not more than -105dB (UNweighted) ref 0FS. This, no doubt, is due to practical acoustic loudness hearing range under best NC we can build included (120 dB max and 25dB noise floor acoustically at spec freq)). Doing the numbers again, as before, we find levels will still be well within audibility.
These are from a 24 bit 48KHz sampling rate and at 997Hz and without jitter 'noise' added into the numbers for dynamic range. Lower freq test for THD+N could be worse as well. Lower bits and different sampling rate systems etal will degrade the numbers well enough for many people to hear differences. It isnt best case -- all theory. And, it isnt worst case but a middle ground Standard.
So what starts out appearing like plenty good enough ... isnt. More needs to be done and can and will be. But, direct from masters is a real major step up.
Addendum..... I would add that the THD number (without the +N) at the -60db level is the more important number and not at 0FS number. It has major impact on details which are important to sound being more 'real'. it is this area where analog has had a traditional advantage... as its thd decreases with level decrease. We need more bits to do that or some cleaver tricks applied.
THx-RNMarsh
The HEADLNE specs are given in order of importance (or perceived importance). here is what I mean by Headlines :
View attachment headlines spec.pdf
Notice the Dynamic range is listed first in importance. And the number is -118dB. Lets look a little closer to that number.
View attachment Detailed spec.pdf
We see it is: 1. the weighted number. 2. It is based on mfr definition of 0 FS which is 3 v rms in this case. It is basically the thermal noise floor to max output (near clipping).
But that is not the Standard value. In fact the standard does include THD and Noise and IS the actual dynamic range to compare according to the IASA. In this case, it is 'typically' -90 dB re 0FS as shown in specs. See Standard here:
View attachment ADC Dynamic range standard.pdf
And, if we use a output level more in line with home audio -- 1 volt rms we can subtract up to 10db from these numbers. That brings the listener to 'typical' - 80dB dynamic range. And, no minimum spec is given to which up to 10db would also be subtracted for home apps.
To these realistic numbers and normalized to the standard.... It seems to me very likely it is above the audible threshold.
BUT, wait. There is more. I am comparing the ADC standard to a DAC spec.... a standard I would like to use because with Hi Res download from master which used ADC we could get exact copy.
But the Standard is also realistic on that DAC side as requiring not more than -105dB (UNweighted) ref 0FS. This, no doubt, is due to practical acoustic loudness hearing range under best NC we can build included (120 dB max and 25dB noise floor acoustically at spec freq)). Doing the numbers again, as before, we find levels will still be well within audibility.
These are from a 24 bit 48KHz sampling rate and at 997Hz and without jitter 'noise' added into the numbers for dynamic range. Lower freq test for THD+N could be worse as well. Lower bits and different sampling rate systems etal will degrade the numbers well enough for many people to hear differences. It isnt best case -- all theory. And, it isnt worst case but a middle ground Standard.
So what starts out appearing like plenty good enough ... isnt. More needs to be done and can and will be. But, direct from masters is a real major step up.
Addendum..... I would add that the THD number (without the +N) at the -60db level is the more important number and not at 0FS number. It has major impact on details which are important to sound being more 'real'. it is this area where analog has had a traditional advantage... as its thd decreases with level decrease. We need more bits to do that or some cleaver tricks applied.
THx-RNMarsh
Last edited:
Problem with digital is that the low level signals are mashed with dither, with analog its thermal noise we can listen through and psycho-acoustically filter out.
Also it makes me wonder how it is with a source where the highest SPL has the lowest distortion. I believe that nature does it quite opposite.
Also it makes me wonder how it is with a source where the highest SPL has the lowest distortion. I believe that nature does it quite opposite.
OK.Thanks. SO thry are alienating some of their most faithful listeners. Not a good business model.....
Dividing an analog signal does not reduce the dynamic range until noise intrudes. You can't make this statement blindly. That is, if an A/D wants to see 3V a 1V source is amplified by a sufficiently low noise amplifier. Conversely the DAC can be used with a 3V full scale output and simply attenuated by 10dB if the system wants a 1V reference level.
Your example can be used in the system this way at no loss of DNR. BTW I recommended a part (I know the designers too), I don't appreciate the reference you chose.
- Status
- Not open for further replies.