Hello everyone! This is my first post so I hope I'm not breaking any rules.
I am looking to build an inexpensive audio compression/limiting circuit using a 3.5mm headphone Jack as the input and 3.5mm for the output as well. I am also looking to make an HDMI version of this as well in the future, where the video signal passes through and only the audio is manipulated. I wanted it to be passive, but I've come to understand that probably isn't possible considering my needs mentioned below. If I could power it with batteries that would be awesome, but micro/usb-c would be acceptable.
For the compression I'm looking for a very high ratio limiter, with fast attack and fast/medium release. Basically my objective is to squash the audio input and normalize it without clipping, the threshold should capture just about every audio input signal, excluding the noise floor (static, etc.).
This is not for music applications, the overall objective is to take whispers and screams that might be in a podcast or something and have it all come out at the same volume level. I want to eliminate any dynamic range from the audio source.
I'm a novice electronic DIY'er with an audio engineering degree, and have arduino's and a couple of component kits, etc. I know I will probably need some special transistors or maybe an IC chip, but information on the Internet is very scattered.
Will I need an internal amplifier? Does anyone know how to work with HDMI signals? More importantly, what IC's or components in specific can I order to start tinkering with the circuit?
Thank you so much I'm advance! I know it is a loaded topic...
I am looking to build an inexpensive audio compression/limiting circuit using a 3.5mm headphone Jack as the input and 3.5mm for the output as well. I am also looking to make an HDMI version of this as well in the future, where the video signal passes through and only the audio is manipulated. I wanted it to be passive, but I've come to understand that probably isn't possible considering my needs mentioned below. If I could power it with batteries that would be awesome, but micro/usb-c would be acceptable.
For the compression I'm looking for a very high ratio limiter, with fast attack and fast/medium release. Basically my objective is to squash the audio input and normalize it without clipping, the threshold should capture just about every audio input signal, excluding the noise floor (static, etc.).
This is not for music applications, the overall objective is to take whispers and screams that might be in a podcast or something and have it all come out at the same volume level. I want to eliminate any dynamic range from the audio source.
I'm a novice electronic DIY'er with an audio engineering degree, and have arduino's and a couple of component kits, etc. I know I will probably need some special transistors or maybe an IC chip, but information on the Internet is very scattered.
Will I need an internal amplifier? Does anyone know how to work with HDMI signals? More importantly, what IC's or components in specific can I order to start tinkering with the circuit?
Thank you so much I'm advance! I know it is a loaded topic...
1) forget passive, what you want is complex.
2) restrict yourself to Audio to make it possible, HDMI signal is a completely different game.
It can certainly be done, there must exist some Audio/Video processor out there doing that and more, but it´s beyond the scope of normal DIYing.
3) have a long hard look at this IC and its capabilities, you can build what you want around it.
There´s others, of course.
https://www.mouser.com/datasheet/2/407/THAT_4316_Datasheet-247985.pdf
2) restrict yourself to Audio to make it possible, HDMI signal is a completely different game.
It can certainly be done, there must exist some Audio/Video processor out there doing that and more, but it´s beyond the scope of normal DIYing.
3) have a long hard look at this IC and its capabilities, you can build what you want around it.
There´s others, of course.
https://www.mouser.com/datasheet/2/407/THAT_4316_Datasheet-247985.pdf
I would start by looking at THAT corporation 4315 chip, it is readily available at low price from most distributors and does have good documentation. It is a popular choice for analog pedals etc; but there are other THAT chips that may be more suited to your need. The HDMI requirement may be sidestepped using a cheap commercial HDMI audio extractor and a matching HDMI audio injector. A completely different solution would be fully software: extract the audio stream from your source material, process with digital filters, then recombine the audio with the video stream if required. The digital option is more versatile and way more effective, the analog one may be easier/faster to setup and integrate with your existing application.
I didnt even know the THAT chip existed. Certainly a good way to go. Another, much more expensive way would be the https://www.hifiberry.com/shop/bundles/hifiberry-elk-audio-bundle/...
The "elk audio" part - https://elk.audio/audio-os/ - is an OS that runs on a rPi and provides a platform for VST plug-ins. One can be pretty sure there's a compressor plug-in, in VST form. There's many others as well available and, as an audio engineer, one would think such a platform would be the cat's meow.
Someone even has cooked up a AI (neural net) application that "listens" to a particular audio device and renders a VST plug-in that - supposedly - emulates the device in software, which then runs on the rPi. So these days you can have your analog tape compression with Dolby noise reduction as a software effect; that's in the realm of actual possibility.
I'll load this topic with that one and pull the trigger. As an audio engineer, why not build a box that can do anything, as long as you're building something? I'm pretty sure, this is the way of the future - (insert hoakey 60's TV commercial echos on the word "future"...)
Just say Eurorack build! Any number of diode or similar circuits will work. Copy Neve. Since this is DIYAUDIO, I will suggest a Nutube based version of a Federal AM-864 compressor from WW2 with something some can sub the remaining pentodes with. It has the synth flavor you want. It would be cool to rethink it, e.g. pentode detector stage, with output NUtubes I have a Dizengoff all tube setup. I wonder if the overall circuit has some merit making it smaller or going JFET, etc. Maybe an 864 8HP build with low voltage subminiature vaccum tubes (socketless). I'm dreaming. Time to build this damn thing....
For logging? Or for listeners with hearing loss?
And FWIW: my hearing-aid only has 3:1 compression in a narrow range, not wideband infinite.
And basic limiting WILL bring up the hiss. A major problem for many users. Noise-gates exist but can be obtrusive.
Nearly all DR work goes better with delay and logic.
And what does "3.5mm" have to do with it? Baseband audio, you can adapt any plug to any hole. (As said, HDMI is different.)
Interesting, I am just committing myself to design a hearing aid for my Brother who´s sick of his ultra modern microminiature in-ear one.
Do you have any technical data on yours or know some dedicated Forum, covering the Tech/electronics/acoustics angle?
Just starting this, no specific time limits, that´s one of the reasons I got interested in THAT VCA ICs.
Starting with a blank piece of paper, as usual, but fully open to suggestions and ideas.
IF you have something, please PM me, so as not to hijack this thread.
Thanks.
When I read PRRs post, I tangentially wondered why DIY of such an instrument isnt more popular? Especially among those in the know of pretty much all things audio? Clearly it would be difficult to touch SOTA "aids" of these days, but since when has that stopped or even paused anyone from their own thinking and attempts at getting good sound from the outside of one's skull into it?
To compensate for mild hearing loss of a relative, I have got good results with a DBX119 unit inserted between the CD output and the amplifier input. This device would also be a solution for the original poster requirement, at least it would fullfill the purpose of testing the kind of performance analog solutions can give.
Modern digital units are better but also more complex; the DBX119 just has two knobs and it works well despite the age. The one I found was sold as "fully working", but of course the original electrolytic capacitors were dried out. It was easy to repair anyway. For modern streaming sources a VST plugin would be the logical solution, but I haven't tested it.
Modern digital units are better but also more complex; the DBX119 just has two knobs and it works well despite the age. The one I found was sold as "fully working", but of course the original electrolytic capacitors were dried out. It was easy to repair anyway. For modern streaming sources a VST plugin would be the logical solution, but I haven't tested it.
Thanks.
That´s about my starting point, : electret capsules (stereo is essential for intelligibility in a reverberant room, against background noise, etc.) > preamp > bandwidth limiting between 250Hz and 8kHz > strongly rising slope between 1kHz and 8kHz : 2:1 (in dB) compression > volume pot > 200mW or so earphone chipamp > in-ear headphones or monitors
Modest requirements, more than happy if my Brother, who I had not seen in years, can reasonably communicate with his better half, last week visit was painful
Not aiming for state of the art micro-miniaturization (which is impossible for me), happy with a cigarette pack size unit which he can put in his shirt pocket or on the table, go figure.
He has a late model in-ear one and speaking frankly, it´s useless.
I am an old style Op Amp guy, so no DSP, too late to teach the old dog new tricks.
But I welcome any data anybody may point to, about hearing aid specific EQ curves, measurements, intelligibility tests, etc.
Maybe service manuals, etc. , if available.
DBX merged into THAT.
Attachments
If you are interested on the way hearing aids works, check out this research paper dated 2011: commercial devices sold today follows the architecture described in the paper. https://ebuah.uah.es/dspace/handle/10017/15361
It also contains a short hearing aid history with pictures and description. By the way, the transparent mode of some wireless noise cancelling headsets is actually a "poor's man" hearing aid function, maybe with less background noise than a old fashioned analog device. Your proposed 8KHz bandwith is probably too wide, less than 5KHz is more likely to be the hearing limit of someone with impaired speech comprehension.
Last edited:
Way back in the early 1990's, when hearing aid signal processing was still analogue and I did my ingenieurs (MSc) thesis on hearing aid compressors, they had two compressors:
2:1 compression starting from an adjustable, relatively low input signal threshold
Infinite compression starting from a certain, relatively high output level
The first compressor was meant to improve intelligibility, the second compressor was meant to prevent further hearing damage. There was also an adjustable filter in the signal path and I think there was a hard clipper at the output.
My thesis is in Dutch and my present computer can't read the floppy discs on which it is stored (or any floppy discs for that matter). I must have a hard copy somewhere.
As you probably know, besides dynamic range compression, boosting the upper mid by a couple of decibels is a well-known trick to improve intelligibility. See the frequency response of the Sennheiser MD-21 for example.
2:1 compression starting from an adjustable, relatively low input signal threshold
Infinite compression starting from a certain, relatively high output level
The first compressor was meant to improve intelligibility, the second compressor was meant to prevent further hearing damage. There was also an adjustable filter in the signal path and I think there was a hard clipper at the output.
My thesis is in Dutch and my present computer can't read the floppy discs on which it is stored (or any floppy discs for that matter). I must have a hard copy somewhere.
As you probably know, besides dynamic range compression, boosting the upper mid by a couple of decibels is a well-known trick to improve intelligibility. See the frequency response of the Sennheiser MD-21 for example.
Last edited: