Jump to content
IGNORED

Upsampling strategy - genre dependent?


Recommended Posts

Hi. Much discussion about Upsampling comes down to preference of Linear vs. Minimum Phase filters (pre vs. post ringing etc..)

 

I am wondering, is there any generally understood "rule of thumb" when Upsampling from 44.1 to 96 with Jazz vs. HipHop?

 

In other words, is one Upsampling method clearly superior to another when the track is genre A vs. genre B?

 

thanks for any feedback

--jay

Link to comment

I personally have the same preference across genres. Miska has discussed different filters per genre for HQPlayer, but I don't recall whether the differences were a matter of minimum vs. linear phase.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
I personally have the same preference across genres. Miska has discussed different filters per genre for HQPlayer, but I don't recall whether the differences were a matter of minimum vs. linear phase.

 

hoping Miska can chime in!

Link to comment

Having just spent a lot of time with HQP, after having read many hours of threads on the subject, the most important thing is what sounds good to you. There are so many permutations of systems, including the big unknown of how the filters, upsampling and dithering of your DAC interact with HQP's. And do not forget to go back to you prior baseline to see whether it really does sound better to you. Reading things here should not be the deciding factor.

Link to comment

For me the upsampling strategy depends entirely on the kind of filter that was used in recording/mastering the CD. If it was recorded with a half-band filter (down 6dB at 22.05 kHz but with aliasing so that the total energy in each band does not seem to have any rolloff) then I try to find the lowest frequency with significant aliasing (typically 20-21.5 kHz) and use that for my stopband. If it was recorded with a gentle rolloff to a silent guard band above 21.5 kHz, then I try to respect the recording/mastering engineer's choice of a good-sounding rolloff and preserve the content by using a steep, phase-linear 22 kHz filter.

Link to comment

Not univocal answer: what resampling algorithm better for ciertain genre.

 

Here possibly watch from physical point of view.

 

As example, acoustic solo guitar or piano music.

 

Minimal phase filter

One side:

- smooth front,

- more artefacts after real front.

Other side:

- no sound before real front. Here it's important thing due rarefied sound.

 

Linear phase filter

One side:

- Front better,

- lower level of artefacts after real front.

Other side:

- sound before real front.

 

However it depend on certain filter.

If degree of nonlinearity minimal phase filter is low, remains only higher level of artefacts after real front.

 

For replete arrangement, me seems, no difference.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment

It's conceivable that your ears would be more or less averse to certain artifacts depending on the instruments or genres. I myself have selected filters based on what sounds good to me in the most challenging genres (chamber music for strings and French opera), and find that the same filters sound good in all genres. My ears are strongly averse to any measurable degree of aliasing (even around 24 kHz), moderately averse to both pre- and post-ringing, and relatively indifferent to a loss of bandwidth above 18 kHz or so. I do not find minimum-phase or other apodizing filters beneficial; I prefer linear-phase, as do most other users on this forum who have expressed a preference. Indeed, I'm not quite sure who has a strong preference for minimum-phase filters.

 

Here is the impulse response for what I use most often (sox rate -v -s 384000 sinc -a 60 -t 3000 -20400):

Screen Shot 2015-05-09 at 11.21.10 PM.png

Compared to many other commonly used filters, the first few side peaks are typical, but the ringing more than 0.3 ms from the main peak is greatly reduced.

 

Here's the impulse response for an apodizing filter (sox rate -v -a -b 91 -p 25 384000), which corresponds to the foobar plugin setting that I suggested to jaynyc as closely corresponding to the type of filter he requested in the other thread.

Screen Shot 2015-05-09 at 10.58.27 PM.png

Notice how much more postringing it adds compared to the rather modest amount of preringing it removes.

Link to comment

Re preference for minimum phase:

 

- Charles Hansen (principal of Ayre)

 

- Miska (developer of HQPlayer, designer of DSC1 DAC)

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Duplicate

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Notice how much more postringing it adds compared to the rather modest amount of preringing it removes.

 

Exactly! Right pictures. It's conservation of energy.

 

When we calculate filter without ringing, we calculate half of FIR (impulse response) array (after front only).

 

However for achieving suppression after stop band like full length FIR array, need calculate half of FIR twice (both after front input signal).

 

Thus we get "double ringing" after the front input signal.

 

General energy of ringing is not (almost is not) changed.

 

Due the conversation of energy matter "linear vs min. phase" is univocal as I said before.

 

If we will use intermediate phase, total energy will anyway distribute between "pre-" and "post-".

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...