Hey everyone,
This is my 2nd year as a member of the Audio Engineering Society (aes.org) and so far I am happy with the papers and journals I am able to get through my membership.
Still I am wondering if there is an alternative source for scientific papers in the field of DSP for Audio? Where is cutting-edge work usually first published? What sources do you follow with?
Cheers!
Back at university I found the IEEE quite resourceful. But to be honest, as an indie developer I didn’t invest in a membership…
Thanks @daniel. IEEE membership is similar as cost (as much as I can see). I wonder - how do their libraries compare specifically for papers on DSP in Audio?
IEEE peer review is much more strict than the j. AES, at least recently (probably due to the thousands of papers they had to toss a few years back). The Transactions on Signal Processing and Signal Processing Magazine are also cited far more than the j.AES, but part of that is probably bias because they don’t focus on audio and are published more frequently.
Worth noting though that the IEEE papers are not really geared towards applications in pro-audio (it’s basically everything else, from automotive to defense). AES and DAFX are still the go-to’s for pro audio. That said, there’s a lot of DSP out there that is worth learning about outside of audio that the j.AES and DAFX don’t carry.
And this is anecdotal, but algorithms more applicable in commercial/consumer products like AEC, VAD, homomorphic filtering of speech, LMS/other adaptive filtering, etc doesn’t get published as often in the AES or to as high a quality as the IEEE. I’ve also had difficulties when authors take things invented for speech processing in telecoms (phase-vocoder, LPCC, etc) and try and put a pro-audio spin on it in the j.AES whereas the IEEE articles are much better for understanding the motivations behind a technique. But that’s just my opinion.
Thanks so much @Holy_City! I didn’t even know about the DAFX papers db.
Perhaps still a tiny source (volume-wise) but seemingly at the forefront of Audio + Machine Learning - Google’s DeepMind team is likely worth the mention in this thread. They seems to become a great source for new ideas and invention and their WaveNet paper and the demonstrated results were really impressive.
Hi Nikolay! 
