A question about Fantom X's playback engine

[compusic]

First, I have searched this forum, but still confused.

My question is ---- is Fantom X's playback engine hardwared with deemphasis function? That means all the samples in the ROM and expansion cards are pre-emphasised?

So if for better result, I must pre-emphasis other samples, created by myself, before imported into Fantom X?

And as for Roland's own "Fantom-X Sample Converter" tool, I don't have to worry about emphasis, right? Since samples on S7X series CDs are pre-emphasised, and the tool does not have the deemphasis function. Or the tool does deemphasis in the background? if my guess is wrong.

BTW, any other formats of samples, of hardware samplers, have the pre-emophasis built-in?

Thanks in advancer for any help!

[Artemiy]

Yes, you are correct, all internal waveforms are pre-emphasized and they are de-emphasized on playback. As for S samplers, not sure, but I guess the technology is the same.

[compusic]

Thanks for reply.

Then next I'll guess the reason for a problem you have discussed in another post ( since you and others seem to have no mention to this reason) . That's why the resampling has inferior quality.

I'm afraid when we listen to the result by resampling, the file has been deemphasised twice. First time is while playing back the origional sample, Fantom will deemphasis it and then records ( or resamples) the deemphasised result. Later when play back the recorded result, Fantom will deemphasis once more, since the deemphpsis function is hardwared. Is it right?

[Artemiy]

Yes, if you resample twice, de-emphasis will happen twice too. This is why you need to pre-emphasize after each resampling.

[Andy Keys]

How would you do that, Art?

Andy

[Artemiy]

Andy, the pre-emphasis function is in the Sample Modify menu (in the same place with trim, normalize, etc.).

Note that:

- You need to pre-emphasize after each resampling procedure if you want to use the sample inside the Fantom.

- You DO NOT need to pre-emphasize if you will transfer this sample to your computer.

[Andy Keys]

Thanks, Art. That's very useful to know. )

Andy

[dbijoux]

Thanks also for the clarification on samples heading to the computer. This thread is quite easier to make sense of.

If I have this correct, the sound engine deemphasizes the signal for output, which includes resampling. ie:

[table]
[tr][td]Sound Engine[/td][td]--->[/td][td]Output circuit[/td][td]--->[/td][td]Deemphasis[/td][td]--->[/td][td]Output A/B[/td][/tr]
[tr][td][/td][td][/td][td][/td][td][/td][td]Resampling[/td][td]<---^[/td][td][/td][/tr]
[/table]

[wudsiba]

This is something I've never quite been able to figure out. I know to normalize after resampling, but why is it set up like this?

[TheMvuser]

No No No..

don't Normalize and then emphasize.. in fact when you resample make sure you do it at -12db. The emphasis EQ curve is a 12 db boost starting at around 4000k. if you normalize anything with a lot of high frequencies (like a Hi Hat) and then add emphasis you will clip the highs badly. These procedures I am reading on this forum do not take into account WHAT the pre-emphasis curve might even look like.

Record a swept sinewave at -12db from a computer oscillator signal generator.. load that into your machine.. add emphasis to it.. Transfer it back and take a look at it on a sample editor.. THEN you will see the Curve.. it is a BOOOST... if you boost a normalized signal you get ----- >> Distortion

[dbijoux]

Who said normalize?

Now you're talking my language, I graphed a comparison of the resampling de-emphasis vs pre-emphasis vs a low-pass filter.

One thing you should clarify - it's the pre-emphasis EQ curve.

These procedures I am reading on this forum do not take into account WHAT the pre-emphasis curve might even look like.

(see above)

I'll give you the bit on the normalize before preemphasis, this has been covered as well - in not so many words.

[TheMvuser]

Re:
Quote. wudsiba
This is something I've never quite been able to figure out. I know to normalize after resampling, but why is it set up like this?

you shouldn't normalize after sampling and (then) add pre emphasis.
if you are going to normalize you need to do it after.

Roland need to put in an algorithm that allows people to alter gain.
otherwise you would have to waist tons of time making sure that every sample is NOT at normalization. beeecause, you are going to have to add pre emphasis to EVERY sample you import to get the system operating at design spec frequency response.

some Roland machines have emphasis and some have pre-emphasis as the terminology. they are the same thing really.

not sure which Rolands have de-emphasis active on the outs and which don't.
I have heard some people complain about Fantom G being too bright.
If the G has it switched off on the outs for some reason, that could explain why their libraries sound bright. because those Libraries are pre-emphasized.
to compensate for that you could apply a master EQ with a negative -6db high shelve starting at 5k. .. db Frequency responses alter in ratio to Frequencies btw.
so the db ratios are different at different frequencies. That is sort of why a +6db at 5K can make a 12db difference at 20K. it is also dependent on the actual EQ architecture you are using. <-- this is a Fantom G I am refering to btw. I am not saying this even is the case for the Fantom G. I don't know, I don't have one.

btw... by the time the frequency response of a swept sinewave hits 20k it is at +12db in my tests.

-13db is the general minimum safe scale for any sample, form what I can see.