File-Based Audio Done Right at John's

[nat8808]

For the terminology, is upscaled meaning bit rate, as opposed to upsampled?

[MartinT]

I've only used upsampled: in my case I took a 16/44 CD rip and used SOX to upsample it to 24/96 with noise-shaped dither. This is proper upsampling since noise has been added to hide any quantisation effects.

The effect we heard must be down to the replay chain pushing the brick-wall filter higher up and out of the audio range, but I must say I was really surprised at the positive improvement.

[nat8808]

I thought I'd heard the term 'upscaling' before, but I think it only applies to video..

That then made me think that upsampling only meant the sample rate and there was another term for the bit rate..

I wonder if it might have sounded better at 88.2 instead of 96 - no rounding errors to entertain with a straight doubling of samplerate.
How something handles complicated maths of say 44.1 into 96 can be strongly dependent on the competance of the algorythm and whoever wrote it. Whether one hears the difference much between good and bad algorythms I don't know - I remember reading Tom(?) Faulkner saying he did (the audio engineer who wrote in Hifi News from time to time).

[John]

It was a surprise to both of us but on both tracks the effects were obvious and is certainly worth investingating

[MartinT]

Have you got SOX working yet, John?

[MartinT]

Nat - good point about 88.2, I shall experiment with it. I can also push up to 192 to see what effect it has.

[John]

I give it another try tonight Cheers for your help

[NRG]

176.4 I would have thought Martin not 192...

[Gazjam]

When I ran a squeezebox setup I played around with SOX up sampling and found it to be a bit of a mixed bag tbh.

Coming from the point if view of objective listening, rather than just going by the numbers, I found many factors could determine wether it was a positive improvement or not.

Whether you used your computer or your dac to upsample gave different "results" using the same Setting....even the the style of music using the same settings determined wether it was an improvement it not IMO.
It's a very subjective field...different SOX algorythyms do change the sound, but which one do you prefer...and does it work in YOUR system?

We all like to quantify things, the best this, best at, but my findings were like a lot of other areas in this game, highly subjective.
It's free to try and you may like it, give it a go.
After much gnashing and grinding of teeth, lashings of FUD and 3 months of my life I'll never get back, what was best for me was trying to get the best original file source I could, things like not ripping to Flac, or even Wav..but Aiff because to me it SOUNDED better when it entered my lug holes. source first, as Ivor T used to say.

The golden rule for me is as little processing and extra overhead as possible.
Get the original file sorted first, if its a favourite album, buy a better quality pressing, have the option of a 24/96 download? Get that instead.

With my Touch setup I really looked in to this in some depth.
Never used Linux so much for years!
Editing the convert.conf comparing different pre and post echo settings, linear phase vs minimum phase vs polyphase, allow/disallow anti aliasing, dither on/off, how MUCH volume attenuation to apply and did it affect sound quality rather than just avoid clipping...all that stuff.

Yup, it did change the sound but my findings were in the end that it was analogous to running a graphic equaliser, but a lot more subtle and a helluva lot more work.
The level of return for me was not worth the effort and high levels of FUD in the long run .
I even found there to be a "best setting" for close miked acoustic tracks, a best setting for classical, best setting for rock...that kind of thing.
Maybe it's a stage I had to go through to get back to where I was and make the most of what I had.

Coming back to the PC arena, one thing I did settle on (when I was playing around with upsampling settings on Foobar and Jriver) was that upsampling 16/44 to 88.2 sounded better across the board than upping it to 96.
This was the start of my going down the "less is more, get the source right" way of thinking.

Is free to try and it may work for you in your system with your ears.
Like everything in is hobby you should try it for yourself - especially if its free.

[MartinT]

Thanks Gaz and Neal. Most interesting. I'm definitely going to play with 88.2 upsamples. 176.4 is not an option as the Bushmaster doesn't decode it. Also, looking at the time it takes for SOX to upsample a track, there is no way that on-the-fly upsampling is going to give as good results.

I'm slightly puzzled at the WAV reference, since this is the native representation of a file with no compression or other handling. It always represents the file as-is.