Peaks easter egg sample location

I’ve had a look through the github project and cannot for the life of me work out where the samples for the special modes could be. I’d like to try and replace the vocal with a noise sample to get a crazy noise generator, as beautiful as her voice is I think longer term an aggressive noise like sample could be really useful for stepped percussion with the slicer that’s in the easter egg.

The other one I’m going to look at trying is removing the morse background noises too and having the 4 controls vary pitch of the root on a trigger. As a mini bleep box.

Any help in this would be fantastic.



make -f peaks/makefile resources to convert it into the big array in

Then you need to remove the XOR masking here if you want your data to be played verbatim (there’s some basic XOR encoding to make the sample harder to locate in a raw .wav or .bin firmware data file).

great! I think I’ll get cracking tonight once I’ve finished this abysmal daytime coding I do :slight_smile:

Sounds like a great idea! Please report back with your progress.

Cool - would love to hear how this turns out.

Finally had some time to start looking into this at long last! Stupid sexy wage paying clients …

Anyway, I’m looking at the digits.bin file and was wondering if there is a rough spec sample wise I should work to … Mono obviously, but more size etc … as I’ve been unable to use Audacity to extract it so far into a playable item.

Also on removing the encoding should the line look more like this in the end?

int32_t a = static_cast<int32_t>(wav_digits[integral]);
int32_t b = static_cast<int32_t>(wav_digits[integral + 1]);

Sorry for what is probably a pretty low level question. I should really just buy a drum machine and be done :smiley:

Sorry to bump this again, I’ve been in the number station code again and for the life of me can’t figure out how I should prep a sample for it. Is there a bit rate / length that the existing one is that you can reveal at all? I’ve managed to figure out how to modify the distortion to be a little more controllable as well as the noise chain, but that damn sample keeps beating me!

I’ve also tried importing the .bin in many different way in audacity as a raw file

As Olivier notes above, the voice sample has been subject to lightweight XOR encryption, where each sample has been XORed with a transformation of the phase accumulator value corresponding to that sample. Thus, if you listen to the data file, it just sounds like noise. Your best bet to to remove the decryption code entirely and just substitute unencrypted sample data. I know almost nothing about digital audio sampling, but it is probably just a mono 8-bit PCM sample at the sample rate that Peaks runs at (48kHz, IIRC).

No, not at a 48kHz same rate, otherwise it could only store sames of total duration of one second or less. More likely an intentionally very lo-fi 4kHz sample rate or similar.

I’ll give that a shot.

I’ve been looking at the various raw export options in Audacity, I expect changing .raw to .bin would suffice.

However there is a multitude of options surrounding the raw exports. Which is where I’m stumbling.

The encoding type, i.e 8bit signed / unsigned PCM / VOX … etc …

The headers I expect should be simple RAW

for note a 4khz amen break in mono is 307kb so 10x the size of the existing .bin

BennelongBicyclist any idea how the size of the .wav translates to flash size on the chipset and what the limits are, I know from bees in trees you are the master working to the limits :wink: hence the reduced chars even.

Pichenettes if you were able to privately share the file that has the XOR encoding removed to analyse the sample for length etc that would be fantastic.

8-bit unsigned, no header, 6kHz. In this format the file takes, 6kb per second.

The person who was recorded for the sample does not approve of its release in an unencrypted form. I don’t really see what this would bring anyway - The voice_digits array contains the offsets (in #samples) of each “subsample”.

I really don’t think it’s a good idea to try to shoehorn a sample player into this.

Thanks …

I’m not trying to shoehorn anything too much. Either a single break or much better high hats. I use peaks for drums 99% of the time and the 808 style closed hat is a little weak in comparison to the rest of the drums, especially the bite of the FM mode kicks and tom type sounds. So I will probably add something like a Linn hat, which in combination with the ringmod / noise and chops will be great.

Thus turning my case into a needlessly expensive groovebox. I’d love to see a Roland Boss group meeting, with a sales manager trying to get a 9u modular approved on their inventory :slight_smile:

What about just rewriting the code from scratch then?

It’s on the list, but as you can see by the time since this thread first popped up … time to sit and learn how to has been very limited.
Also the thought of coding after a day of it has been getting less and less ( 6 months of a very abrasive client ) I may even manage to get a vacation at some point! It’s been about 10 months.

I am going to order a STM32F4 discovery board today and start working from there to learn the fundamentals calls of the libraries. Out of interest is there any variants on the current boards out there you would recommend over others?

“Clouds” or “Elements” (or later this year: “Warps”) are great platforms for developing audio processing code :slight_smile:

Just as well I have an elements sitting happily in a case right now :slight_smile: Which I was looking at the code for only last night.

When I asked Jamil, my colleague from Lahore, for his help in identifying the language spoken by the mysterious Peaks numbers station broadcaster, I couldn’t help feeling that after listening to her intently over and over, he had fallen just a little bit in love with her. Or maybe that was just projection on my part. Anyway, it is only appropriate that she remain modestly encrypted under a filmy veil made from the finest XOR.

Is Warps using a M3 or M4?

M4 - this module is more CPU intensive than Elements!