[Amberlamps]

don't forget Bandcamp !

Whose going to be playing the flute ?

 

[JaZZ]

For Rob and anyone who can answer this.

I have a strange question i was hoping for an answer. if the latency is reduced for video with mscaler with the video filter then how come we don't hear a delay with 1M taps processing just audio? How is this compensated for? i understand the delay for audio is 0.71 seconds and for video 0.65 seconds.

"Yes it creates a lot of delay as its using 1.42 S of music (44.1) to do the interpolation, giving a 0.71 S of delay. The delay for 48k (video) is 0.65 seconds, hence the need for low latency (low delay) filter option".

I haven't used the video mode yet, but according to Rob's explanation as to the differences in signal processing the HMS upsamples to 768 kHz in video mode as well, but uses a simpler WTA filter for reduced delay – thus clearly less than 1 million taps, I would think. Rob's above quote looks strange, since 0.71 s to 0.64 s is barely a difference worth of sacrificing optimal sound quality. Probably it should mean «...and for video it's using 0.65 s of music/sound, resulting in a delay of 0.325 s».

 

[musickid]

So how come with 1M taps and just audio we do not hear a delay? What is the mechanism at work here?

 

[Siz.]

As triode says, once you near and get used to robs mscaler, although it might not be apparent at first, the more you listen the more you will hear things that you didn!t hear previously, timing and seperation is the thing I noticed most.

Everything seems to have it’s own space instead of being all jumbled together.

If you are thinking about buying one, go for it, as it’s something that will be around for many years to come. dacs can come and go, but the mscaler will still be relevant 5+ years from now.

Top tip for any mscaler and headphone user, Look for the Chesky label.

The Chesky brothers are probably the main people who are making hires binaural / meant for headphone users, recordings.

I have a few Chesky made albums from HDtracks, Amber Rubarth, Macy Gray stripped. Those albums can be bought from hdtracks.

Thanks for this. Having read through this thread from the start, I can see a compelling narrative around just how transformative this bit of kit is; testimony such as yours has more than convinced me to audition it (and start buying the wife some flowers!)

But I am also aware of reports of less than stellar performance improvements in some cases, and these interest me; I am just trying to bottom these out.

I am not particularly interested in the pass-through discussions: it may be that the inclusion of a pass-through button, while well intentioned, has back-fired somewhat: the nature of the improvement in SQ expected from this type of device is unlikely to jump out in a quick-fire A/B test - except to the designer, who knows exactly what he is looking for. The rest of us may better appreciated it if we stop pushing buttons, stop ‘Listening’, switch off and just let the music in. An A/B switch may be unhelpful in this regard.

But I am interested in the reports of users sometimes not experiencing a big improvement from HMS. It doesn’t seem to be a golden ears thing; rather it seems to be a sporadic phenomenon affecting certain tracks only - which I have argued might be explained by sub-optimal processing of at the ADC stage.

If this is true then it is quite important as it may have implications both in terms of assessing the M scaler’s performance and fo M scaler owners in sourcing music.

But the theory needs to be tested against the experience of M scaler users - and in particular those who have experienced variable performance enhancements.

Apologies for the length of this post - I am not sure if any of it makes sense to anyone, but as always I live in hope . And thanks for the Chesky tip!

 

[Triode User]

Thanks for this. Having read through this thread from the start, I can see a compelling narrative around just how transformative this bit of kit is; testimony such as yours has more than convinced me to audition it (and start buying the wife some flowers!)

But I am also aware of reports of less than stellar performance improvements in some cases, and these interest me; I am just trying to bottom these out.

I am not particularly interested in the pass-through discussions: it may be that the inclusion of a pass-through button, while well intentioned, has back-fired somewhat: the nature of the improvement in SQ expected from this type of device is unlikely to jump out in a quick-fire A/B test - except to the designer, who knows exactly what he is looking for. The rest of us may better appreciated it if we stop pushing buttons, stop ‘Listening’, switch off and just let the music in. An A/B switch may be unhelpful in this regard.

But I am interested in the reports of users sometimes not experiencing a big improvement from HMS. It doesn’t seem to be a golden ears thing; rather it seems to be a sporadic phenomenon affecting certain tracks only - which I have argued might be explained by sub-optimal processing of at the ADC stage.

If this is true then it is quite important as it may have implications both in terms of assessing the M scaler’s performance and fo M scaler owners in sourcing music.

But the theory needs to be tested against the experience of M scaler users - and in particular those who have experienced variable performance enhancements.

Apologies for the length of this post - I am not sure if any of it makes sense to anyone, but as always I live in hope . And thanks for the Chesky tip!

Personally I think you are overthinking it.

 

[mlxx]

I haven't used the video mode yet, but according to Rob's explanation as to the differences in signal processing the HMS upsamples to 768 kHz in video mode as well, but uses a simpler WTA filter for reduced delay – thus clearly less than 1 million taps, I would think. Rob's above quote looks strange, since 0.71 s to 0.64 s is barely a difference worth of sacrificing optimal sound quality. Probably it should mean «...and for video it's using 0.65 s of music/sound, resulting in a delay of 0.325 s».

I think he meant the delay for 44.1k is 0.71s and for 48k it is 0.65s thus a video mode with less delay is required.

So how come with 1M taps and just audio we do not hear a delay? What is the mechanism at work here?

The delay is still there. There are just no moving lips to see things are out of sync.
There is a delayed start to the music and it stays constant that way.

 

[musickid]

There is a delayed start to the music and it stays constant that way?? Could you explain this please? surely if there was a delayed start then when you press play we would have to wait for 0.71s for the music to start on roon for eg? i'm not hearing this delay. many thanks mk. also roon has it's own settings to ensure gapless music AFAIK.

 

[mlxx]

There is a delayed start to the music and it stays constant that way?? Could you explain this please? surely if there was a delayed start then when you press play we would have to wait for 0.71s for the music to start on roon for eg? i'm not hearing this delay. many thanks mk. also roon has it's own settings to ensure gapless music AFAIK.

Gapless, there is no start/ stop at all so same as music just playing continuous.
Pausing/Resuming the music should just continue on.
Starting new music with new sample rate will have a start delay.

 

[jbrownson]

Before I got my HMS when I listened to DSD I switched my Dave to DSD mode instead of PCM mode. Now I'm going through HMS first. I believe HMS always outputs PCM anyway, so there's no sense in switching Dave's mode right? How does the processing of DSD differ when going through M Scaler to Dave's PCM vs Dave's DSD mode? Is it purely better to go through HMS vs Dave's DSD mode? Maybe a question for @Rob Watts ?

 

[ZappaMan]

There is a delayed start to the music and it stays constant that way?? Could you explain this please? surely if there was a delayed start then when you press play we would have to wait for 0.71s for the music to start on roon for eg? i'm not hearing this delay. many thanks mk. also roon has it's own settings to ensure gapless music AFAIK.

You’d have to imagine an Input buffer inside mscaler that holds a stream of audio 2.3 secs long, before the tip of the stream leaves mscaler to hit the dac. The stream will always be 2.3 secs long after it leaves your source.
Any buffering that goes on in the source is separate.

 

[musickid]

So in real world terms is this delay audible only when you first start playing music with any one particular SR? if i switch to another SR say 2 or 3 hours later you get another delay with audio but because the delay is so small you don't really notice it? i just need to know when this delay (or delays occur) and how often to round this up. many thanks. i'm learning everyday as they say. doesn't feel right owning something worth 3.5k and not understanding the basics.

 

[mlxx]

So in real world terms is this delay audible only when you first start playing music with any one particular SR? if i switch to another SR say 2 or 3 hours later you get another delay with audio but because the delay is so small you don't really notice it? i just need to know when this delay (or delays occur) and how often to round this up. many thanks. i'm learning everyday as they say. doesn't feel right owning something worth 3.5k and not understanding the basics.

Correct.

 

[musickid]

with the delay is the first 0.71sec of a track in effect missing or does it finish 0.71sec late due to starting late? i understand this not for every track just the first one for every particular SR??

how do you know all this mixx?

 

[ZappaMan]

with the delay is the first 0.71sec of a track in effect missing or does it finish 0.71sec late due to starting late? i understand this not for every track just the first one for every particular SR??

how do you know all this mixx?

Every time you change the instruction to your source, say switch out an album, then mscaler will need to ingest 2.3 (or however long, performs its maths), before the dac gets sight of it.
That which starts late, will end late.

 

[musickid]

so you never actually miss any part of a track (no matter how small) it will always play completely from start to end? it's just the time delay which affects the data going from the source to the mscaler then finally to the dac?