POKEY Intonation

[BoatofCar]

I have always thought that in some instances notes sounded flat through POKEY, particularly the M.U.L.E. theme. Listening to the ANTIC interview with Bill Hogue yesterday, he confirmed that this had to do with the nature of the chip itself. Is there an article/post on here where this is talked about in more detail? I tried searching but didn't get anywhere other than just references to POKEY music sounding out of tune.

[+Nezgar]

It's simply because the POKEY, by default, uses an 8-bit frequency divider and those resulting 256 'tones' do not all line up well with the frequency intervals of the chromatic scale.

 

Many other sound chips including the SID in the C64, and the YM2149 in the Atari ST use 16 bit frequency dividers, giving 65536 tones, so it was much easier to generate frequencies matching notes on the chromatic scale.

 

The POKEY can be configured to use 16 bit frequency dividers as well by combining two channels into 1, for a maximum 2 16-bit channels. (or 4 with 2 POKEYs), but this is too much of a sacrifice for most music you hear on the Atari 8-bits.

[R0ger]

The main issue is pokey frequency registers are 8 bit. That is too small resolution. On clean square wave you get 2 decent octaves, after that it's getting dicey.

Then there are basses. Typically $Cx distortion is used for those. That uses 4 bit polynomial to generate deeper sounds, which generate 15 bit long sequences, which are then sampled by the pokey generator. Usually 2 sets of frequency values are used, each giving different timber, and different frequency range of the resulting tone.

One is named C bass in RMT. These values have to be divisible by 3, but not by 5. The resulting tone has 5 times lower frequency than the pure square tone. So now instead of 256 values you only have 55.

The next bass is called E in RMT. These values must not be divisible by 3 or 5. That's 8 values out of 15. Slightly better than C bass. But the resulting tone is 15 times lower than pure square tone. So the bottom octave, which is the most precise, is useless, it's too deep.

All that can be solved by merging 2 channels together, which will get you 16 bit frequency resolution. But it limits amount of channels drastically.

 

Most of the time, it's the bass tones which sound off, while the lead tones sound more or less OK. There is no final solution to this.

There are some attempts to get better bass, like discussed here:

https://atariage.com/forums/topic/292342-new-bass-approach-rediscovered-in-old-song/

 

or here:

https://atariage.com/forums/topic/272971-hard-released-a-new-demo-at-sv2017-reharden/

 

 

 

Edited February 24, 2020 by R0ger

[emkay]

Actually, we should discuss the term of "flat sounding".

The real issue that POKEY sounds mostly rather flat is that there a squarewaves only . They always produce the high peak "corner" sound.  Other Chips produce triangle, or sawtooth by definition. SID has the low pass filter in addition, to cut any corny noise...

 

There are tricks available, to have deeper sounds with reduced high noise.

 

One is to use 15kHz with mixed frequencies.

 

 

 

 

POKEY also has some feature that no emulation is using. It seems to have an "volume overload" mode that does exactly what is needed to cut unwanted high sounds.   

 

 

 

 

[R0ger]

I wouldn't listen to emkay. He's recent experiment demonstrate quite clearly he is 100% tone deaf. If not completely deaf.

[emkay]

2 minutes ago, R0ger said:

I wouldn't listen to emkay. He's recent experiment demonstrate quite clearly he is 100% tone deaf. If not completely deaf.

Writes Who exactly?

[BoatofCar]

That makes sense. I was wondering if it had to do with actual acoustical properties in temperament, if such a thing exists with synthesized sound. 

[emkay]

 

21 minutes ago, BoatofCar said:

That makes sense. I was wondering if it had to do with actual acoustical properties in temperament, if such a thing exists with synthesized sound. 

 

If you combine channels, POKEY creates sometimes unexpected results.

Particular the spread of correct notation frequencies changes. That's why music creation tools were needed to adjust that.

Particular RMT misses Tracker features. This would cause to create thousands of instruments, to have clean notes.

 

Have a look at the wave creation here:

 

 

 

 

 

 

 

And have a listen for some enhanced programming (just like Trackers allow to program PAULA on the Amiga)...

 

 

 

 

 

 

[R0ger]

Listening to some tunes I found one more factor .. detune effect is often used on Atari.

In simple form you play the same note on 2 channels, where 1 of the channels is shifted 1 value up or down. The resulting sound has interesting new timbre, but sounds slightly above or bellow the original note. You could solve it by using 1 bellow on one channel, and 1 above on the second channel, but that already gives you 2 values difference, the timbre is different, and it's not used that often. The same trick is then used in what RMT calls 'filter' sounds, which use POKEY's so called 'high-pass filter'. The timbre is different, but the problem with slight frequency shift remains.

Usually problems like these can be easily masked with slight vibrato, and it's actually main reason vibrato is used all the time on Atari

[antrykot]

1 hour ago, Nezgar said:

 

 

Many other sound chips including the SID in the C64, and the YM2149 in the Atari ST use 16 bit frequency dividers, giving 65536 tones, so it was much easier to generate frequencies matching notes on the chromatic scale.

SID does not use 16-bit frequency dividers, it has 24-bit phase accumulators.

[ivop]

20 minutes ago, R0ger said:

Usually problems like these can be easily masked with slight vibrato, and it's actually main reason vibrato is used all the time on Atari

Even in 15kHz mode it helps. I used it on the bass in Procrastinate.

 

[emkay]

17 minutes ago, antrykot said:

SID does not use 16-bit frequency dividers, it has 24-bit phase accumulators.

How about :

 

Fout = (Fn*Fclk/16777216)Hz

 

and on the C64 (1MHz clock)

 

 

Fout = (Fn*0,0596)Hz

[ivop]

19 minutes ago, antrykot said:

SID does not use 16-bit frequency dividers, it has 24-bit phase accumulators.

True. Internally. Jeri Elsworth made a YouTube video explaining it:

 

 

The 16-bit register is what she draws at 1:09

 

[Poison]

1 hour ago, emkay said:

Writes Who exactly?

You are totally deaf Many of your music is totally false But you can not hear it

[emkay]

10 minutes ago, ivop said:

True. Internally. Jeri Elsworth made a YouTube video explaining it:

 

 

The 16-bit register is what she draws at 1:09

 

The notation correctness is at 16 Bit. Even better to have clean notes played , without any step between the frequencies, as real Synthesizers do. And then to have the chance to do some tuning on the notes, if needed. As particular music isn't working correctly , if everything is 100% in tune.  

[emkay]

3 minutes ago, Poison said:

You are totally deaf Many of your music is totally false But you can not hear it

If you start to make music on the Atari, I'd possibly believe your words. 

[Poison]

5 minutes ago, emkay said:

If you start to make music on the Atari, I'd possibly believe your words. 

Yes you're right. I'm not as good at making music as you are. I'm sorry, but hopefully I can still improve. I hope to reach your level soon and the cats playing the piano

[+Stephen]

3 minutes ago, Poison said:

Yes you're right. I'm not as good at making music as you are. I'm sorry, but hopefully I can still improve. I hope to reach your level soon and the cats playing the piano

Here's another great example of out of tune:

 

[emkay]

8 minutes ago, Poison said:

Yes you're right. I'm not as good at making music as you are. I'm sorry, but hopefully I can still improve. I hope to reach your level soon and the cats playing the piano

hehe...  getting polemic again?

I never claimed to be a musician, and never would do a "Soundcloud" channel, because I KNOW, it will never sound like music.

 

Even funnier the link that Stephen posted to add to you post, as it is exaclty pointing to something to think of.

If the AI of Youtube finds the correct key that is keeping a "real" piece of music, why do you think it is out of tune, even after a software is calculating the dependencies ? Do you think they are all deaf? Even the calculations?

[+Stephen]

53 minutes ago, emkay said:

hehe...  getting polemic again?

I never claimed to be a musician, and never would do a "Soundcloud" channel, because I KNOW, it will never sound like music.

 

Even funnier the link that Stephen posted to add to you post, as it is exaclty pointing to something to think of.

If the AI of Youtube finds the correct key that is keeping a "real" piece of music, why do you think it is out of tune, even after a software is calculating the dependencies ? Do you think they are all deaf? Even the calculations?

If you think that is in tune, you are completely deaf.  There's no point arguing it.  It's like arguing with someone that screams "the sky is orange not blue".  How can you convince them otherwise?  Why bother?

[Marius]

Omg guys, stop with that 'deaf' insult. It makes no sense. Be honest, so the discussion here is: who can create the music that is the least out of pitch on pokey?

 

Is this 'being deaf'actually a real argument? Almost all pokey music is terrible when it comes to pitch accuracy; yes poison your music too. 

 

From a musician's point of view I see you guys trolling about who is the most deaf. Well let me give the redeeming word. When it comes to pitch accuracy, everybody should better be deaf to be prevented hearing all this pokey-crap. What a nonsense; it really makes me angry.

 

Pretty funny to find a Poison demo called I am Blind (which perhaps better could have had another title), whilst the music is also not that pitch accurate (listen at 0:36-0:48, 1:03, 1:34, 1:38 or 1:45 for instance, omg what were you thinking; not my idea of pitch accuracy, or being in tune in what way?)

I do not blame you at all poison; this is exact what is pokey-music is all about. But you should be way more cautious calling other people deaf. It does not make any sense at all. Listen to your own stuff, and conclude.

 

 

 

 

Or listen to your 'No Good' song on soundcloud (excellent title again). The first 38 seconds or so, there is definitely some intonation-problem going on there. The distance between bass and 'melody' is too large and exposing the intonation problem: there is definitely some 'floating' going on in the frequency/pitch. That is not only a flaw of pokey, but also in your arrangement... your choice as a musician. I personally do not care, but from a musician's point of view it sounds horrible. 

I am absolutely not after bashing you. Not at all. But you found it needed to disqualify emkay as being deaf, and then I thought: hmm the pot is calling the kettle black. 

 

Edited February 25, 2020 by Marius

[ggn]

1 hour ago, Marius said:

Or listen to your 'No Good' song on soundcloud (excellent title again). The first 38 seconds or so, there is definitely some intonation-problem going on there.

Then again, many would argue that Prodigy do produce music altogether so that part of the argument might be void there...

[Poison]

1 hour ago, Marius said:

mg guys, stop with that 'deaf' insult. It makes no sense. Be honest, so the discussion here is: who can create the music that is the least out of pitch on pokey?

Is this 'being deaf'actually a real argument? Almost all pokey music is terrible when it comes to pitch accuracy; yes poison your music too. 

From a musician's point of view I see you guys trolling about who is the most deaf. Well let me give the redeeming word. When it comes to pitch accuracy, everybody should better be deaf to be prevented hearing all this pokey-crap. What a nonsense; it really makes me angry.

Pretty funny to find a Poison demo called I am Blind (which perhaps better could have had another title), whilst the music is also not that pitch accurate (listen at 0:36-0:48, 1:03, 1:34, 1:38 or 1:45 for instance, omg what were you thinking; not my idea of pitch accuracy, or being in tune in what way?)

I do not blame you at all poison; this is exact what is pokey-music is all about. But you should be way more cautious calling other people deaf. It does not make any sense at all. Listen to your own stuff, and conclude.

Or listen to your 'No Good' song on soundcloud (excellent title again). The first 38 seconds or so, there is definitely some intonation-problem going on there. The distance between bass and 'melody' is too large and exposing the intonation problem: there is definitely some 'floating' going on in the frequency/pitch. That is not only a flaw of pokey, but also in your arrangement... your choice as a musician. I personally do not care, but from a musician's point of view it sounds horrible. 
I am absolutely not after bashing you. Not at all. But you found it needed to disqualify emkay as being deaf, and then I thought: hmm the pot is calling the kettle black.

You are right, may be, but try to listen it in HQ versions on Bandcamp

 

https://poison6502.bandcamp.com/track/iam-blind

 

https://poison6502.bandcamp.com/track/prodigy-no-good-start-the-dance

 

Soundcloud has terrible noisy soundquality, i do not know why. And Iam blind YT video, is not my official product it is only badly recorded music and put together vith some video

 

My music may not be quite right, but emkay's songs are sometimes really extreme and no matter what great filters or techniques he uses, it sounds awful

 

[thorfdbg]

1 hour ago, Marius said:

Omg guys, stop with that 'deaf' insult. It makes no sense.

 

Well, unfortunately, it does. emkay's music *is* out of tune pretty much all of the time...

 

However, concerning techniques: It is not strictly necessary to use 16 bit counters and hence waste one channel. The resolution is sort-of OK for the lower tunes, but it becomes a problem for higher-pitched notes. What one can do instead is to use a base-frequency of 1.79Mhz for the higher notes, and get away with an 8-bit counter, thus switch AUDCTRL dynamically. Unfortunately, this only allows to extend the resolution of two channels, not all four of them. For lower frequences, AUDC=Cx works - sort of. POKEY is more a sound effect generator than a music synthesizer, though.

[R0ger]

Being tone deaf is not an insult, it's blessing as far as pokey music goes. And being completely deaf is a blessing as far as Emkay's music goes.