Various dither options are built into most programs such as Reaper. In computers , usually they use triangular probability distribution to approximate Gaussian, as it is easy to calculate by averaging random numbers.
For final downsampling when further resampling is not expected, most people including me typically used one of the shaped dither options to move noise out of the frequency range of higher hearing sensitivity.
Regarding dither bit depth, that's usually fixed and not user adjustable.
By default in most audio programs, some dither is always applied when downsampling. Someone would have to go out of their way to defeat or disable it, if it was unwanted for some reason.
I will take a look at the link you provided, thanks.
EDIT: I would add that when I have experimented with different dither options, I haven't found found much difference. It seems very small. To my reckoning, it wouldn't seem to account for the differences I have heard. Also would add that 24-bit sound cards are usually specified as adding some dither to the analog audio before A/D conversion when recording at 16-bits. When recording at 24-bits, system noise is sufficient to effectively provide dither. That's what they say.
Last edited:
Ah ok, from what you had written wasn't clear if you just brute forced to redbook which would of course lose things.
I don't suppose you have any samples of this effect you could post? would be very interesting to see if we can spot this and also (and possibly more interesting) if you can train us to spot it.
I don't suppose you have any samples of this effect you could post? would be very interesting to see if we can spot this and also (and possibly more interesting) if you can train us to spot it.
TPDF dither actually is the theoretical best fit to perfect linear interpolation between bits leaving a flat noise floor. Noise shaping moves the noise to less audible places and the floor is no longer flat.
From my measurements vinyl has at best about 6 LSB's of gaussian noise which would obviate the need for dither. I would think all but the most extreme microphones would also not need dither at the original recording end. Bob Katz has a fairly level headed discussion with respect to dither and processed/mixed audio
I might be able to find some files, but I would need a place to post them. Wave files are impractical to post here.
EDIT: The full project files I would need may be on some old disk drive or maybe on one of my backup drives, I will have to look. I at least I know off the top of my head where a nice mp3 encoded directly from 24/96 is. Maybe a 24 bit wave file too. If so, I could downsample it.
Last edited:
Okay, I stand corrected on TPDF. Agreed about noise shaping, I didn't mean to imply otherwise.
Even though some recording sources may not need dither, some 24-bit ADC cards or boxes still apply dither noise to the analog signal before conversion when recording at 16-bits.
That's area I need to revisit, now that I have much better performing replay gear.
In my experience with mixdown from 24 bit to 16 bit I found that I preferred no dither.
Adding dither put a constant white noise under the music, and different dither options added other 'flavouring' to the overall music, none of which I liked.
Dan.
Well, I'm not having much luck finding old 24-bit projects. I could post an old mp3 from 2004 that I recorded if anybody wants to listen to music.
However, for our present purposes, I have a new 24-bit sample of a tambourine in stereo. One channel is a Schoeps CMC-641 and the other is a Neumann TLM-103. Both are going into Grace 101 preamp variants, which are pretty clean. Then into the HEDD ADC at 24/96 (clean - no effects processing enabled). From there I made a few different 16/44.1 wav file versions. They vary in terms of resampling algorithms and dither.
I'm guessing most people can probably hear differences between at least some of the files.
However, usually a lot of the disagreeable stuff is above 8 - 10 kHz. If not sure how good hearing is up in that region, there are free software online tone generators out there that could be used to make quick test tones.
Also, to hear the background noise and dither, good hearing sensitivity for quiet sounds might be helpful (tinnitus might be a hindrance, for example).
Because there are several files, ftp seems like it might be preferable to bitorrent, unless maybe I package it all up in a big zip file?
However, for our present purposes, I have a new 24-bit sample of a tambourine in stereo. One channel is a Schoeps CMC-641 and the other is a Neumann TLM-103. Both are going into Grace 101 preamp variants, which are pretty clean. Then into the HEDD ADC at 24/96 (clean - no effects processing enabled). From there I made a few different 16/44.1 wav file versions. They vary in terms of resampling algorithms and dither.
I'm guessing most people can probably hear differences between at least some of the files.
However, usually a lot of the disagreeable stuff is above 8 - 10 kHz. If not sure how good hearing is up in that region, there are free software online tone generators out there that could be used to make quick test tones.
Also, to hear the background noise and dither, good hearing sensitivity for quiet sounds might be helpful (tinnitus might be a hindrance, for example).
Because there are several files, ftp seems like it might be preferable to bitorrent, unless maybe I package it all up in a big zip file?
Zip file to Dropbox or the like would work.
Post a public link and anybody can download it.
Dan.
Post a public link and anybody can download it.
Dan.
Dang that's me out. Oddly my tinnitus had ebbed to being unnoticable during the last week (off work) and after one day back in the office I am sitting here and, whilst quieter than the tick of the kitchen clock, it's noticable in the otherwise silent house.
Anyway I will download and listen tomorrow. And promise no peeking at the files until after I have listened 🙂
Schoeps is on the left, Neumann on the right. The former has a smaller diaphragm and should be smoother at high frequencies. The latter is lower noise. Might be helpful to listen to one at a time.
Also, for some recording programs such as Reaper, if the files are all loaded into one project, care has to be taken that Reaper is running at the correct project sample rate, otherwise it will merrily resample any tracks as needed on the fly without any warning. That could potentially reduce playback sound quality.
Also, for some recording programs such as Reaper, if the files are all loaded into one project, care has to be taken that Reaper is running at the correct project sample rate, otherwise it will merrily resample any tracks as needed on the fly without any warning. That could potentially reduce playback sound quality.
Funny, mine used to be part time, but since my op and all the drugs I am on I now have a less intense but full time background hiss... Hoping it will diminish with time and as my drug intake decreases (one in particular gabapentin).
Downloaded the files will try them on headphones later (main system in bits due to new decoration scheme/domestic dispute!!! one faction says the speakers go with the room one faction (wife) says no (I wonder if she remembers Eastenders Dirty Den era...))
It was the case with my mixer with myriads of 4558. It was cheaper to add myriads of resistors and replace a power supply; it's ground was not well designed, also double regulation helps a lot.
I know that feeling. I am trying to negotiate a set of control-1s to go in for centre channel and surround. Conceptually she has accepted the benefits of this, but the trial fitting will possibly be a contentious affair.
(on topic because the irony in opamp discussions is that, at least for classical music with only 2 channels you are stuffed for fidelity).
I'm not quite sure what we are supposed to be listening for in the tracks Mark posted but I have listened to them all and made a few comments. I haven't opened the files in anything other than Media Player on W10.
Are we to post here or pm with impressions ?
Are we to post here or pm with impressions ?
Marce,
The Gabapentin is about as effective as an aspirin in killing any nerve pain. Perhaps your doctor can take you off that or substitute something else. My mother the RN works in a clinic and they give out Gabapentin like candy, little real effective except for a placebo. I use to give them to my now passed wife as they gave them to her for pain which never worked, I tried them also once for really bad sciatica and they didn't help in the least. Talk to your doctor and see if they can give you something that is really effective and hopefully not addictive such as Oxycontin. From what I understand aspirin in large doses or over long periods is also a cause of Tinnitus, I have a slight case of it but only in one ear, kind of weird, I thought I was hearing the power supply in my desktop computer for a long time until I finally noticed the slight buzz in one ear when the computer was off. Most of the time I don't notice the background sound, I've been trying to figure out how I can identify the frequency the Tinnitus is at?
The Gabapentin is about as effective as an aspirin in killing any nerve pain. Perhaps your doctor can take you off that or substitute something else. My mother the RN works in a clinic and they give out Gabapentin like candy, little real effective except for a placebo. I use to give them to my now passed wife as they gave them to her for pain which never worked, I tried them also once for really bad sciatica and they didn't help in the least. Talk to your doctor and see if they can give you something that is really effective and hopefully not addictive such as Oxycontin. From what I understand aspirin in large doses or over long periods is also a cause of Tinnitus, I have a slight case of it but only in one ear, kind of weird, I thought I was hearing the power supply in my desktop computer for a long time until I finally noticed the slight buzz in one ear when the computer was off. Most of the time I don't notice the background sound, I've been trying to figure out how I can identify the frequency the Tinnitus is at?
- Status
- Not open for further replies.