Bit depth change - Resample Required?

[kwecter]

Hi,

The way I understand it, here is the order of the settings I'm using. Have I got it right?

bit depth 32 float (reference 13)
resample 96000
volume normalize peak
volume normalize fixed amplification if needed (no replaygain on playback)
equalizer if needed
replay gain
bit depth minimum 8 bit maximum 24 tpdf (or bit depth 24)
resample 44100 (or 48000 for 24 bit)
playlist writer

The original files are 16/44 wav, I'm converting to iTunes for a car. Trying to get the best sound possible. Are the resamples necessary? Is the order correct? Comments welcomed ...

[Spoon]

Re: Bit depth change - Resample Required?

>resample 96000

Not required, and the 44100 if keeping the same frequency.

[kwecter]

Re: Bit depth change - Resample Required?

>resample 96000

Not required, and the 44100 if keeping the same frequency.

Is this setup totally lossless throughout? Or lossy somewhere?

[Spoon]

Re: Bit depth change - Resample Required?

If you change the audio, then is not lossless (volume normalize peak changes the audio).

[kwecter]

Re: Bit depth change - Resample Required?

If you change the audio, then is not lossless (volume normalize peak changes the audio).

If you were converting wav 16/44 to iTunes for the car and wanted the best quality possible, What settings in which order would you use? No ReplayGain on the playback and iTunes is limited to 24/48.

[garym]

Re: Bit depth change - Resample Required?

when you say itunes, do you mean playable on an idevice? If so, then ALAC (apple lossless is the best). Converting 16/44.1 to 24/96 or higher doesn't magically turn the CD quality files into hires files. But for a car, lossless is way overkill. For my idevices I maintain a mirror lossy library created from my FLAC files. I use mp3 (lame) at VBR -V2 (~192kbs).

[kwecter]

Re: Bit depth change - Resample Required?

when you say itunes, do you mean playable on an idevice? If so, then ALAC (apple lossless is the best). Converting 16/44.1 to 24/96 or higher doesn't magically turn the CD quality files into hires files. But for a car, lossless is way overkill. For my idevices I maintain a mirror lossy library created from my FLAC files. I use mp3 (lame) at VBR -V2 (~192kbs).

Yes I'm converting from wav to Apple lossless and playing it on an iPod jack in my car. The wavs are 16 bit, but they seem to "sound" better going down to 24 bit rather than 16/tpdf. Bigger file? Since my car speakers are my best speakers right now, I'm trying to get the absolute best sound possible. If not lossless, then lossy - whatever "sounds" best.

[Spoon]

Re: Bit depth change - Resample Required?

You cannot increase the quality of a 16 bit file by converting to 24 bit, the extra information is not there.

[harjac]

Re: Bit depth change - Resample Required?

Hi Spoon,

Can we rip to higher resolution like 24/92 with Batchripper in Flac ??

[Spoon]

Re: Bit depth change - Resample Required?

There is no point in doing so, you do not get extra quality by changing the bit depth or sample rate.

[garym]

Re: Bit depth change - Resample Required?

are these 16/44.2 rips from CDs? I assume you understand that converting these to 24/192 is equivalent to adding a bunch of zeros. Larger files with same info. That is, 0.00000000 instead of 0.00000.

[Dat Ei]

Re: Bit depth change - Resample Required?

That is, 0.00000000 instead of 0.00000.

Not quiet the same...

From a mathematical point of view 0.00000000 and 0.00000 are different because the precision of a number is +/- the half of the last digit. So the precision of 0.00000000 is +/- 0.000000005, and the one of 0.00000 is only 0.000005.

A standard Audio CD offers only 2^16 different discrete values. If you transform them distinctly into a range of values coded with 24 bit, you still have only 2^16 different values. You could invent some mathematical methods and models to compute interim values to fill up the range of values (i.e. Interpolation), but the result would be a guess and not the pure signal.

You could do the same for the time axis (i.e. 192kHz instead of 44.1kHz), but that would be a guessing model too. It might work with some songs, but it might fail with others.

It's a little bit like upscaling digital Pictures.


Dat Ei

[harjac]

Re: Bit depth change - Resample Required?

From Tested.com I quote:

But How Does It Sound?
24-bit sound is a tricky thing to gauge. Does it provide for a greater resolution of sound? Definitively. It has room for 256 times the data, remember. Are you going to be able to hear that difference? Harder to judge. Human hearing supposedly tops out at 20kHz, but that doesn&*8217;t make higher sample rates useless. According to the Nyquist rate, to fully capture a wave, it should be sampled at twice its highest frequency. In other words, a higher sample rate, and a greater bit depth, gives your sound more wiggle room, meaning sound peaks are less likely to be truncated and the subtleties of the music are less likely to be drowned out.

So there is a benefit... isn't there ?
If not why Hires files ?

[garym]

Re: Bit depth change - Resample Required?

good article on 24/192 files:

http://people.xiph.org/~xiphmont/demo/neil-young.html