Stereo Tool

New IMAGE file with BETA055 installed, including PiMP with 4.2 ms latency! (Pi 4 only)
https://www.stereotool.com/download/Thimeo_Pi34.zip

This image starts up with MicroMPX, you can switch to another product such as PiMP via System Info:
Please note: There are changes in the image file, you cannot use the previous image file with the new installers (well you can, but you'll get audio hiccups in PiMP).

What is the recommended way to connect a mic to PiMP and get the processed audio back out?

You need a pre-amplifier, possibly with phantom power (depending on the type of microphone). The HifiBerry sound cards expect normal line in levels.

(Small other update: For non-multiband processor, or if you don't want to have any frequency splits below 1000 Hz, we can probably get the latency down to slightly under 3 ms. It's added to the code now, but untested. It actually might sound ok for the existing microphone presets, they typically treat everything below about 3000 Hz as a single band.).

Note: There are still some issues with which thread runs on which CPU core on the latest version that I posted. I'll try to fix that over the weekend. The even lower latency (2.9 ms) doesn't run properly yet, because the CPU usage it too high - 3.2 ms might be ok but I want to try to fix the CPU load problem.

Updated installer!
https://www.stereotool.com/download/Thi ... 92-060.zip

PiMP can now go down to 2.85 ms, but loading or browsing through the web interface can occasionally cause a small audio glitch. The CPU load is still a bit higher than at higher latency settings, probably due to framework and thread synchronization overhead stuff, and recovering from glitches is a lot harder at these extremely low latencies. (There might still be some settings that we can tweak to improve the behavior).

To run PiMP at 2.85 ms, set the "Latency" setting on the I/O page to 64 samples instead of 128. To get rid of most of the glitches, increase the output buffer size to 0.3 ms. That adds 0.33 ms of latency (3.18 ms in total).

Edit: Good news! It turns out that glitch recovery mainly has a problem with low frequency tones, which are normally not present in microphone input anyway. I have been testing for glitches by playing a 30 Hz tone - after increasing it to 300 Hz I see no unrecovered glitches in a 3-hour measurement, with the latency at 2.85 ms! So... it looks like PiMP can run at 2.85 ms. I do hear some effect in the audio for which I need to search for the cause at that low latency though; not sure yet if it will actually be noticeable in speech.

Latency results:
PiMP:
- At 128, it runs at a latency of 4.17 ms, with no unrecovered glitches in 24 hours.
- At 64, it runs at a latency of 2.84 ms, but with lots of background noise triggered by low frequency sounds, and frequent glitches if you play low frequency tones and have the web interface open. To prevent this, set the buffer to 0.2 ms, that will increase the latency by 0.33 ms to 3.17 ms, but it sounds MUCH better. With the web interface closed, the default settings should be fine.

STudio:
- At 128, STudio requires 0.67 ms of extra latency to not have constant background noise triggered by low frequency sounds. If the web interface is enabled, it needs to be even higher (1.00). I'm changing the default to 0.5 (which translates to 0.67). This results in a total latency of 4.84 ms. If you prefer a bit of noise over latency, you can lower it.

There is some audible tone at latency 64 if you play a low frequency sound. I'm guessing that this won't really be a problem if you feed it speech.

Dear Hans -

I need to say this here in public. This is simply amazing stuff. Period.

To achieve these latencies in user space applications on a generic Linux kernel requires some very clever coding. Hats off, and please keep us updated. I'll be among your customers when it's commercially available.

Updated installer!
https://www.stereotool.com/download/Thi ... 92-062.zip

We have added a new product: MicroMPX AM Reprocessor. This has a very specific purpose: Lots of stations in the US need to run both FM and AM, where the AM is basically what gives them the right to run on FM, but they need to transmit the same content, and 99% of listeners will be listening to the FM signal. So they would need 2 separate streaming methods: MicroMPX for FM and something else for AM. This Reprocessor takes the mono output from MicroMPX, lowpass-filters it, boosts the level a bit (since all the highs and stereo are gone) and then clips it - asyemmtrically if needed. It's clearly not the best possible solution - but it's very simple to set up. (If there's a lot of interest in this we might at some point create a MicroAM codec, that is explicitly NOT what this is). Because of the clipping, the CPU usage is higher than that of MicroMPX or MicroMPX+, but it should still work fine on a Pi 3 or 4.

Updated installer:
https://www.stereotool.com/download/Thi ... 92-067.zip

Released! See https://www.thimeo.com/how-to-use-the-r ... ard-image/

Next step is to add licensing for the different products to the site, that will hopefully be ready later today.