I’ve never quite cracked this on the DT (MK1). When I first got it, I had assumed that SSR would do this, but it doesn’t, it just makes things sound robotic.
Has anyone found any decent way of reducing the quality of vocals to sound like it’s phone-grade audio? You can sometimes get close with the right mix of EQ and filters, but I’ve not nailed it. Curious if anyone has a go-to technique here.
Digitakt or Digitakt II? There is great info on frequency values for both as they relate to parameter values here: Digitakt Filter Cheat sheet
Telephone should be in the range of 300 Hz to 3 kHz. If not achievable with the EQ filter type, I think a low/high shelf filter would be the best choice, but those don’t exist on the Digitakt (yet?). I know they are an option on the Syntakt analog tracks. For other flavorings I would lean on bit rate reduction instead of sample rate reduction, sprinkles of overdrive and rebalancing with AMP VOL if you want some distortion or megaphone territory.
Honestly though, I think the simplest solution here is resampling (meaning achieved outside the Digitakt).
The easiest way is to compress the audio down to a low bitrate MP3. There are a bunch of expensive CODECs for telecom that get the bitrate lower (which means more calls over a link, which means more profit for the telecom), but the easiest way to get access to them is to use something like Skype (soon to be RIP) either long distance or over a low bandwidth network.
TL;DR: it isn’t SRR you want, but low bitrate audio codecs, specifically those designed (and licensed $$$$) for telecom use.
Thanks both. Ah so it sounds like this is mostly done outside the Digitakt, which is a shame for someone trying to avoid computers but I guess you can’t do everything on one box (though I’ll try the tricks @dzyndzel metions).
And I should have specified that I’m on DT1 (edited post.)
The original Digitakt doesn’t have BRR, nor the option to overdrive post filter (which I would also levy here). So yeah, I think your best bet would be running through a plugin and sampling that or trying what @obscurerobot mentioned. The only other thing I would add RE: .MP3 is that Digitakt / Transfer is going to convert anyway and to keep that in mind.
You might also be able to find someone here willing to do the effect for you, but I imagine you would prefer the flexibility of doing it yourself to taste and for whatever you want now or in the future.
The way I would avoid using a computer would be via a cell phone. Either hold the phone near a speaker or run an audio cable into the phone. Some VOIP services have a test mode where you can call a number or hit a button and “speak” at the system and have it echoed back, to test call quality. Perfect for transforming your audio.
One of the stock plugins on the mpc has a bunch of preset filter effects with settings like bullhorn, earbuds, guitar amp, transistor radio, telephone, vintage cellular phone etc. Since it’s part of the included software, it might also be part of the free mpc daw which would save some time and effort if you just want to convert your audio outside of the dt. I think that the particular plugin is called air flavor but you’d have to confirm.
If you want to do it inside the dt, it might be worth trying some resampling, particularly pitching the sample up maybe half an octave or so and then back down again. Ricky Tinez did some old videos showing one way to do this and exposing some of thhe pitfalls.
There are definitely easier ways to do it but if it were me, I’d find an example of some phone audio that I wanted to emulate and start by identifying in which range and which frequencies are most obviously lost from the phone recording, and immediately try to pull some of that out of my sample, then I’d start resampling and see where that got me.
You’ll probably have better luck if you have an example of what you want which you can A/B against your progress.
The voice memo app on my phone also has a low quality setting, if yours does as well, you might consider rerecording your sample with the phone mic at low quality and start processing it in the DT from there to save some time, because it will immediately cut a lot of the correct frequencies.
Another thing to try if you’re working with short chunks of vocals: run it into the delay at ~100 feedback, turn off the sample after the first pass, filter the HP/LP as it sounds good, then resample the delay after it’s resampled itself a bunch of times. Setting tempo to 30 gives a sizeable delay buffer.
I am a bit perplexed. Given this:
Doesn’t that mean you can get a decent result with just the base-width filter ?
I thought the same but I wonder if the slope isn’t steep enough to truly replicate it???
Dunno
As someone who used to make sound files for telephony systems, the key is in the sample format used by such systems to begin with. They commonly use a 8kHz sampling rate, and some systems also employ mu-law or a-law companding to the sound (some exotic cases used ADPCM). To get such a sound, you reduce sample rate and bit depth of the signal, before doing that you remove the lowest low frequencies.
Now I dont use a digitakt, never have, but as long as it has a BPF + sample rate reduction, you should be able to get there
It’s a sampler….
Use the filter to get rid of the low end.
Resample.
Use the filter again to shave off the high end.
Resample.
Repeat until it sounds good. Maybe distort a little, bit crush or what ever you think sounds nice.
Fuck it Resample that and filter it again.
You don’t have to do it all in one go. Resampling is boss.