File-Based Audio Done Right at John's

[John]

I think its to do with the coding
But to be honest I do not have Scooby do (Clue)

[MartinT]

I don't know why SOX and Audacity encoding should sound different. I'll have to experiment some more.

[electric beach]

Bloody hell - it was a sensible question

All up and about early aren't we! Hands off cocks and on with Sox!

[MartinT]

Always in the office at 6am, Steve. Now for my 2nd coffee...

[John]

I am usually chatting with Fe

[AlexM]

Well, I've upsampled a file in SOX to 32/88.2 and, playing back using VLC, it confirms the file format and the Caiman plays it! First impressions are good so I'll listen some more tomorrow and figure out a way of playing the file on my big system (so far, all my trials have been on the PC system).

SOX command to upsample to 32/88.2:

sox.exe -S -G --multi-threaded %1 -b 32 %2 rate -v 88200 dither -S

where %1 is the infile and %2 the outfile (works as a batch file).

Martin/All,

You don't need to apply dither when upsampling, only when downsampling. Dither is applied to reduce noise in the LSBs of the bit depth-reduced sample when you are averaging values below the 16th bit LSB. You are just increasing noise in the upsampled file by adding dither.

I think that most of the apparent increase in SQ when upsampling red book audio is to do with the differing effect of digital filters applied higher in the frequency range, with less aliasing/time smearing effects on the audible frequency content of the signal compared to the effects of filters when the sample rate is so close to audible frequencies (as it is with 44.1Khz sampling).

There isn't any more HF content other than some noise created by upsampling process, but the effect of the digital filters is moved further away from the audio band. Similarly, increasing the bit depth of a 16bit signal is just padding it with 8 or even 16 bits of zeros, no changes in dynamic range by increasing the bit depth of the source material. It may have an effect on the DAC, depending on whether it truncates bits the 8 LSBs in a 32bit container.

Cheers,
Alex



Cheers,
Alex

[MartinT]

Hi Alex
I do understand, however I was experimenting with adding noise as it represents the least evil compared with quantisation noise. Covering it up (albeit at extremely low levels) seems to benefit sound in the upsamples. However, I shall experiment a little more with and without dither.

[NRG]

If you look at the spectrogram of a 44.1 to 88.2Khz up sample at the same bit depth you'll see some very strange things the two files are quite different.



---------------------------------
Posted using Tapatalk

[John]

I find upsampling a mixed bag On my more heavier material I actually do not like what it does I think its something you have to thread with caution

[WOStantonCS100]

Forgive the diversion; but, in just reading the title of this thread, something smacked me in the head. I've been listening to file-based audio ever since I purchased my first CD. Analog aside (which I'll never give up for many reasons) and considering that all I'm doing, when I record on the Tascam, is creating files, I don't see why I should have an aversion to playing files whether it be from a disc (DVD) or a hard drive. My only real criteria for file playing would be a quiet delivery system and that seems quite doable, if less tactile and visually appealing. But, for the sake of the music... hmmmm... something for me to think about.