r8Brain Pro settings

[Arksun]

I realise there's been a few SRC threads in the past, but I wanted to focus specifically on R8Brain Pro here and its settings.

For those that use the R8Brain Pro, which is your personal preference in regards to the settings.

I was reading a post on voxengo forum from a guy that described the linear phase mode as squeezed, a little harsher than the orig, but more highs, and described the minimum phase mode as a little coloured but otherwise ok.

This was both with ultra steep mode enabled, which I'd always just assumed would be the better option.

But he also mentioned that using it in Linear phase mode, but with ultra steep turned OFF, sounded much better than linear phase with ultra turned on, despite the increased rolloff on the highs.

Has anyone else felt this might be a better setting, and indeed, what is your preferred setting.

So far I've only compared linear and minimum modes with ultra steep enabled and my personal preference is the minimum phase mode.

[Bang]

I always use linear phase with ultra steep mode on. I haven't had any complaints. Its hard in my studio to compare things of different sample rates in a blind A/B test so I ask myself after the conversion is done "does it still sound good?". The answer thus far has always been yes. What I CAN do and oddly haven't done is compare minimum phase vs linear phase RESULTS so that both will be 44.1khz. I'll post them here and we can all listen. Ok fine I'll also do with and without steep mode.

[cooker]

Wouldn't doing the same stuff over and over give an easyer to hear result? I mean if 96khz, convert 44.1 khz, again to 96 then finally 44.1 (or more if you have the time). I use the free one and would like to listen to the difference too.

[Arksun]

After doing some more comparisons today with R8Brain Pro's linear phase mode, I gotta say I'm leaning more towards having the Ultra steep mode switched off, rather than on.

I just find it sounds closer to the original than with ultra steep turned on, as having it on seems to make the transients and top end sparkle a bit stronger than was originally present to my ears (resulting in the soundstage seem a teeny bit tighter/smaller)


When using minimal phase mode the difference between ultra on and off is more subtle, but with linear phase mode, the difference between ultra steep on or off is noticeable enough to concern me that having it on isn't such a great idea.

Could anyone else confirm or agree with this difference of having ultra turned on and off and hazard a guess as to why I'm percieving this difference?.

I'm wondering if this has something to do with the way very steep filters can sound unnatural to our hearing and, even though this is way up at the top end of the audible range, it can still have an affect on how we percieve the entire frequency range.

Doing a frequency analysis seems to confirm this for me. The original 48K .wav had the top end frequencies rolloff off up to the maximum 24khz point. But when looking at the 44K linear phase ultra ON, there was no rolloff, it just suddenly cuts dead at the 22K point.
However when looking at the linear phase ultra steep OFF, it displays a similar rolloff to the original 48K source (albiet at a lower level because it rolls off to 22K instead of 24K)

I've upped the images to demonstrate this:

http://www.arksunmusic.com/images/R8Brain_Comp.jpg

[ISedlacek]

Well, I was also a bit concerned what settings to use for the best results and made some comparisons ... But to be sincere, I found that to my ears the audible difference between these settings is so infinitely small, that it probably does not matter at all ... It was also difficult to tell a sound difference between standard dither and direct conversion from 24 to 16 bit directly via r8Brain ...

So just for fun I keep the minimal phase (because somewhere it was written it sounds more analogue ) and ultra steep mode ...

What I quite miss there is the estimated time of finishing the operation (when you have lot of files to process...)

[RickZ]

This thread coming back to life just now was a big help to me trying to figure out a 'harshness' problem I was working on with loud choral music. I used a relatively short piece with loud and soft parts, and did SRC from 96KHz/32BitFloat to 44.1KHz/16Bit at the 4 possibilities:
1 Linear Ultra
2 Minimum Ultra
3 Linear
4 Minimum

For this test material, option 4 was clearly better, ie cleaner sounding.

And the processing was indeed faster, as the r8brainPRO help information indicates, ie Ultra takes more processing.

Best regards,
Rick Z

[ISedlacek]

Thank you for sharing ... will try it too ...

There is however one strange thing happening invariably when using r8Brain: when you make final tracks in 48 or 96 including limiting to say -0,2dB, after the conversion the level suddenly increases to 0 and slightly above. A bit unwelcome it is ...

[RickZ]

RE: Level Increase with SRC . . I believe the theory behind this has been discussed in other threads, so that leaving a bit of room before SRC is a good idea. In some cases, I've seen the level after SRC creep up a small amount, 0.3db at most. Going back to the test of r8brainPRO mentioned in previous post, I checked the before and after, and in this case one channel was 0.1dB lower (-0.9>-1.0) other channel same (-1.2dB), after SRC. I think it's related to the waveforms, and when peaks hit relative to sample points, so varies from song to song.

FWIW,
Rick Z

[Bob Olhsson]

The average level isn't actually increasing. Different sample rates will sample the waveform at different points and this can often result in peaks that are a few tenths higher. The filters can also ring a bit.

Most DACs resample and filter. Many will distort when you get near full output. Lossy audio coders such as MP3 use filterbanks that will ring and clip anything close to the top.

If quality matters, it's never a good idea to record to full level. Just because you can't hear any problem doesn't mean others using different playback gear won't.