Export Multiple Issue with MP3/Project Rate

[waxcylinder]

... In my case I am also required to submit a copy in that format and bigger files mean bigger credits, so in my case not really a placebo, I actually benefit!

[waxcylinder]

If I record an album at 96 kHz, I am ensuring that I have all the information I can possibly have within the recording itself. If I want any other format in the future I can utilize that file to create the other files ensuring accuracy for those as well.

A good plan, but just be aware that if you create compressed audio files like MP3 or AAC from them, then you will not be retaining "accuracy" as of their nature the compression algorithms for these formats deliberately discard some of the audio data.

WC

[WeirDave]

If I record an album at 96 kHz, I am ensuring that I have all the information I can possibly have within the recording itself. If I want any other format in the future I can utilize that file to create the other files ensuring accuracy for those as well.

A good plan, but just be aware that if you create compressed audio files like MP3 or AAC from them, then you will not be retaining "accuracy" as of their nature the compression algorithms for these formats deliberately discard some of the audio data.

WC

So this begs a question. If I record an album at 96kHz and the same album at 44kHz and I save them both as FLAC, the 96kHz at 24bit FLAC and the 44kHz @ 16bit FLAC. The first question is why are the file sizes going to be different if it is the same information? Also, since they are different what is in the 24bit FLAC that I am losing in th e16bit FLAC? Am I always losing something?

[steve]

Dealing just with the sample rate question (which was what we were originally talking about), look at these two waveforms. What is the difference between them?

tracks000.png (16.52 KiB) Viewed 10991 times

The first has a sample rate of 44100 Hz.
The second has a sample rate of 96000 Hz.
Both tracks have a sine wave that has a frequency of 440 Hz and an amplitude of -1.9 dB.
Both tracks have identical sound. No matter how you test it, there is absolutely no difference between the sound of the first track and the sound of the second. This can be proved by inverting one track and adding the two tracks together - the result is silence (-infinity dB) because the waveforms are identical.
The only difference is that the 96000 Hz sample rate track uses twice as many samples to define the same sound.

[WeirDave]

Dealing just with the sample rate question (which was what we were originally talking about), look at these two waveforms. What is the difference between them?

tracks000.png

The first has a sample rate of 44100 Hz.
The second has a sample rate of 96000 Hz.
Both tracks have a sine wave that has a frequency of 440 Hz and an amplitude of -1.9 dB.
Both tracks have identical sound. No matter how you test it, there is absolutely no difference between the sound of the first track and the sound of the second. This can be proved by inverting one track and adding the two tracks together - the result is silence (-infinity dB) because the waveforms are identical.
The only difference is that the 96000 Hz sample rate track uses twice as many samples to define the same sound.

I understand what you are saying and it makes sense but in your screenshots the 96k file has a smoother waveform so they don't really "look" identical. which confuses me a bit.

[waxcylinder]

So this begs a question. If I record an album at 96kHz and the same album at 44kHz and I save them both as FLAC, the 96kHz at 24bit FLAC and the 44kHz @ 16bit FLAC. The first question is why are the file sizes going to be different if it is the same information? Also, since they are different what is in the 24bit FLAC that I am losing in the 16bit FLAC? Am I always losing something?

I'm no expert here - you're at the bleeding edge of my personal technological understanding

My understanding is that using the higher sample rates gives you more frequent sampling and hence more samples (hence the larger file size) and theoretically more "accurate" digital representation of the analog audio - but remember that one of the jobs of the DAC in your audio player is to smooth the discrete digital samples into as smooth an analog signal to feed to you speakers and your ears - and for most folk a good DAC workimg on a good 44.1 kHz 16-bit signal does a darn good job of this.

I do know from reading other forums that serious archivists do usually use 96kHz 24-bit - but for real-life listening purposes the CD Red Book standard of 44.1 16 is usually fine.

--------------------------------------------------------------

And do remember you are probably losing "quality" all the way down the line:
1) the stamper that was used to press your LP will wear over time - so later pressings are not as "accurate" as the earlier ones
2) you do get cold-pressings (where the stamper is removed too soon) and other bad pressings.
3) The quality of your stylus/tone arm/deck has a big impact
4) the mounting of your deck has an affect too
5) proper alignment and setup of the cart and arm are critical to "quality" retention
6) the quality of the cables and connectors that you use have an impact
7) the soundcard that you use fir digital conversion van have a impact on quality too.

WC

[steve]

My understanding is that using the higher sample rates gives you more frequent sampling and hence more samples (hence the larger file size)

Yes, that's right so far...

and theoretically more "accurate" digital representation of the analog audio

No, that part is not correct.
In the example that I previously posted of a 440 Hz sine wave, whether the sample rate is 44100 Hz or 96000 Hz, there is no difference in accuracy. The sound waves are identical.

The difference is in the highest frequency that can be represented.
There is a theoretical limit of half the sample rate. In practice the limit is a little lower. A sample rate of 44100 Hz can represent frequencies a little beyond the limit of human hearing. 96000 Hz sample rate can represent frequencies way beyond the range of human hearing. For "listening pleasure" there is not really any benefit to reproducing inaudible ultrasonic frequencies, and most good audio systems will filter them out, but as archival copies there may be some benefit for scientific study.

Higher bit formats could be described as producing greater "accuracy". "Inaccuracy" in sample values is perceived as noise of some kind. The type of noise depends on how sample values are "rounded". Conventional mathematical rounding creates harmonic distortion called "quantization noise". An alternative that is often preferred is to randomize the "rounding" with "dither" which replaces quantization noise with a more gentle hiss. The greater the bit depth (bits per sample) the greater the accuracy that is possible (though achieving that accuracy depends on how good the recording/playback equipment is). The more accurate the sample values, the lower the theoretical noise floor. The theoretical noise floor for 24 bit audio is extremely low (lower than even professional audio equipment can achieve).

[Gale Andrews]

When you get close to the maximum frequency a given sample rate can represent (the "Nyquist frequency") you can regard that as very inaccurate:

3800 Hz sine tones at 8000 Hz and 44100 Hz

sshot-8.png (19.21 KiB) Viewed 1778 times

Also, try generating a 3800 Hz sine tone at 22050 Hz (well below the Nyquist frequency) and then a 3800 Hz tone at 384000 Hz underneath, invert one track, select both tracks and Mix and Render. I rendered at 384000 Hz and 44100 Hz project rate. Even apart from the visible glitch at the start of the track, the audio in the middle of the track is not infinity.

Is that result regarded solely as a resampling inaccuracy? Clearly the 3800 Hz tone at 22050 Hz does not look like the 3800 Hz tone at 384000 Hz and it is hard to see how the 22050 Hz track can be regarded as an accurate representation of a sine tone:

3800 Hz sine tones at 22050 Hz and 384000 Hz

sshot-10.png (14.03 KiB) Viewed 1778 times

Gale

[Gale Andrews]

When using the export multiple function after I have ripped an album that I have recorded at 96kHz saving to flac works fine. When you go save as MP3 it prompts you for the Resample rate. That part I am fine with BUT since i am exporting multiple tracks there should be a checkbox or something that allows me to not have to choose the rate for every track.

Wiki Feature Requests has

Invalid sample rate dialogue: Add checkbox "Apply this rate to any remaining conflicts" (checkbox state is stored in .cfg) and/or add checkbox linked to Warnings Preferences "Don't warn again and always use highest rate". (5 votes)

Do you want to add your vote to that, now you know the workaround?


Gale

[WeirDave]

When using the export multiple function after I have ripped an album that I have recorded at 96kHz saving to flac works fine. When you go save as MP3 it prompts you for the Resample rate. That part I am fine with BUT since i am exporting multiple tracks there should be a checkbox or something that allows me to not have to choose the rate for every track.

Wiki Feature Requests has

Invalid sample rate dialogue: Add checkbox "Apply this rate to any remaining conflicts" (checkbox state is stored in .cfg) and/or add checkbox linked to Warnings Preferences "Don't warn again and always use highest rate". (5 votes)

Do you want to add your vote to that, now you know the workaround?


Gale

Sure how do I do that?