Hacker News new | past | comments | ask | show | jobs | submit login
Analog For The People: Synth Master Tatsuya Takahashi On Engineering Fun (npr.org)
144 points by jcrabtr on Dec 23, 2017 | hide | past | favorite | 30 comments



"I called them up and they asked me to come in for an interview — so I brought this thing I'd made, this sequencer synth thing. I showed it to them and they gave me the job."

https://media.npr.org/assets/img/2017/12/08/tatssynth_custom...

This is how you get a good engineering job -- show your impressive thing to people as high up as possible (and everyone else). You'll still have to do a whiteboard interview, but it'll be mostly a formality.


> show your impressive thing to people as high up as possible

How do you even get access to ' people as high up', most programming jobs make you go though "Standardized screening" programming puzzles before they even look at your github, if at all. Many high profile sfo tech companies straight up told me that github is merely a formality in the application and that they don't have time to look at it. Many interviewers get you resume like 5 minutes before the interview and start asking you standard interview questions.


I think the takeaway is that he showed them a complete product, rather than a pile of schematics.

Looking through Github accounts is extremely tedious and boring. On most profiles I’ve seen it‘s hard to tell what all the repos are about, and what the important stuff is.

So if you‘ve made something awesome, you need to market it a bit.

Don‘t point people to your Github account.

Point them to the project website, where they see a screenshot or a demo video, a short, easy to grasp description of your thing, and some impressive stats of your choice that convinces everyone that this project is brilliant.

And make sure the project page also prominently credits you as the author!


There could be a product in there - verified contributions of actual people?


Yes. And I shall call it “my GitHub repo!”


People complain about the coding tests at interviews. I don't think it's as big of a deal as people say. Treat it like a social skill or any of the myriad other unrelated things you are evaluated for at an interview and spend some time practicing.

I think for a reasonable programmer, yes, you are probably going to have to practice your whiteboard interview.[1] But no, it's not that hard, and it's pretty standardized, so you don't have to practice differently at each company you interview.

I personally think that practicing for the programming test is usually way easier than the social test... and I think the programming test is less biased than the social test.

on the other side of the fence, I think it's a good failsafe. Yes, it will fail many qualified people who don't practice, but it will also fail a lot of people who talk a good game but can't code their way out of a wet paper bag.

Overall, I think it's a great counterbalance to how overvalued confidence is at most interviews.

[1]I personally think that you should use one of the screen share programs designed for this; I like https://codepad.remoteinterview.io/ but anything that reduces the amount of handwriting required is going to vastly reduce your false negative ratio without impacting your false positive ratio (unless you code using whiteboards at your company)


> I think it's a good failsafe. Yes, it will fail many qualified people who don't practice, but it will also fail a lot of people who talk a good game but can't code their way out of a wet paper bag.

I don't. I do R&D for a living and have next to no time to relearn those types of undergrad programming techniques because it doesn't apply to the work I do at all. Ask me to create a model for a system, implement a classifier that hits 90% in a way that no one has ever done before and publish a paper on it, I can do that. Ask me about a breadth first search on some tree and I'll look at you blankly. It seems a bit ridiculous that some of my cookie-cutter engineering colleagues(some that relied on me to help them with homework) can make it into places like Google but I can't make it past the 3rd or 4th interview. I can also assure you that none of my colleagues where I work could make it past that same stage of interviews despite having a PhD and despite them being great researchers.


I think I said that it's turning down a bunch of people who are unwilling to practice.

And yes, turning down people who are unwilling to spend a week practicing is a cost... it could be a very big cost if someone else has a intake filtering procedure that allows them to snatch those people up at a lower cost, but it's also a really effective way of filtering out the people who can't learn those things after a few weeks of practice, which is super important.


> unwilling

Can't. I clearly said can't. It's irrelevant to my work and my work consumes all the time I can give to work before I burnout.

> but it's also a really effective way of filtering out the people who can't learn those things after a few weeks of practice, which is super important.

What's your source on that? Can I see the stats that say neglecting more talented engineers so that you don't have to deal with subpar engineers is better than having a larger amount of more talented engineers while having to intermittently deal with subpar engineers? We have brilliant engineers where I work and the average and subpar engineers are in and out inside of a month or two. Didn't seem to hurt us at all in being the leader in our industry.


"Hire easily, fire easily" is another strategy, it's true! but not one used by most places I've worked. It is a slower strategy when sorting through a pool of mostly not good enough candidates, and often makes employees more nervous (Just like how you "Can't" take three weeks off to practice, other people "Can't" get fired after a few months on a new job.)


Out of curiosity, how do I find companies like the one where you work? Your comment resonated pretty strongly with me.


Music equipment corporations are pretty small on the whole, and Korg in particular was originally formed by engineering refugees from a larger gear company, so their culture is a bit more open.


Networking. Work with and impress people who know people, then ask for an introduction if need be. My experience has been that "volunteering" (working for free, if you prefer) has helped me make connections I wouldn't otherwise have the skills/experience/connections to get through conventional means.

Granted, I've actively avoided California, so YMMV.


I remember getting a Korg interview in SV many years ago when I was really into synthesizers and owned quite a collection. The interview was a massive disappointment as it was all about passing a written C++ (feh) test with no room musical passion or domain knowledge - and I was an expert C programmer at the time too.


Any tips or online guides for newbies wishing to pick up electronic music via analog synthesizers? Asking for a friend with no experience and no knowledge of music theory etc ;)


To learn synths, I highly recommend Syntorial: https://www.syntorial.com/. It's $130, but you can try quite a few lessons for free.


Download VCV Rack (it's free and open source, and it has many digital models of real world modules) and start exploring the network of YouTube channels dedicated to analog synthesis. There's currently some kind of Cambrian explosion of small batch analog synth hardware happening, so there's never been a better time to learn. Be warned that the first time you open the app you will not be able to produce sound without help, but that's the same level of user friendliness you'll get from a real modular synth, so it will build character. :)


By far the most helpful resources for me have been product reviews and demos. Specifically the Sonicstate Sonic Lab reviews.

https://youtu.be/eR1PrKfUvsQ

Not really a "guide" like you're asking for, but for me these have been the best because they expose a wide array of products and ways of using them.


Experiment. A Volca is a really cheap way to start, a minibrute is not much more but considerably more flexible. For ~$400 you could pick up and old Virus A, which is DSP-based but has a great sound all the same, especially in unison mode, and is such a well-specified and designed synth that you'll learn a ton of things that could take you several years to pick up on a piecemeal basis.

Of course software solutions abound but there is no substitute for knob twiddling, and doing it in software makes it too easy to look without really listening.

As far as reading material goes, this is the best course available: https://web.archive.org/web/20160403115835/http://www.soundo...


the short answer will disappoint you: pick one you like and experiment.

the long answer: entire genres were built on not knowing music theory or disregarding it (techno, i'm looking at you). most music producers started by turning knobs without knowing what they were doing. most music producers continue to turns knobs just to see what happens, even after learning stuff about synths. the knowledge will come over time ; if you get "the bug" you will inevitably start reading stuff about all this.

of course, there may be exceptions. most "worldwide mainstream" electronic is obeying the usual codes of popular music, so if that's what you want to do, don't listen to me (and maybe produce on a computer).


> the long answer: entire genres were built on not knowing music theory or disregarding it (techno, i'm looking at you).

Thats putting things mildly. The entire electronic music scene was born from experimentation. From the Kraftworks and other synth pioneers, to the beginnings of the dance scene with Chicago house and Detroit techno. From sample heavy genres like Jungle and Hard House, to synth heavy genres like trance. Even IDM was born from producers disregarding popular conventions and look how much that has since bled into the mainstream with tracks incorporating glitchy effects into pop songs.

But it's not just the domain of electronic music either. Prog Rock existed because uni students wanted to push the boundaries. Then Punk came about because they wanted to redefine what it meant to writing music and play an instrument.

So long as music remains an art form, it will be subjected to ignorant hobbyists with a creative flair. And that is what I love most about music; when people say "I have an idea, it might not work but we will have fun trying". But very occasionally, those hobbyists create something so popular they then become trendsetters themselves, like Prodigy, Orbital, Aphex Twin, and the other artists I alluded to above.


You could learn the basics of substractive synthesis (the most common one) for free with software emulations, either on desktop (VST/AU plugins) or with iOS apps (Android is not very well served in this area)

Here's a free barebones synth plug-in (you'll need a VST/AU host, for example Garageband that comes bundled with macOS) : https://tal-software.com/products/tal-noisemaker

To be honest many producers in the younger generation haven't ever touched real hardware synths (let alone real analog synths) and do everything "in the box" using software. There's endless debate on whether this sounds the same (it's the "Emacs vs vi" wars of the synth world), but as Tatsuya says in the last paragraph of the interview, you'd be hard pressed to hear any difference once in a mix.

A few of the best sounding classic-analog-emulating software synths would be : Diva, Serum, SynthMaster, Sylenth1, Spire, and the Arturia collection.

Hardware synths are fun though, and most importantly their interface can spark inspiration where technically equivalent but mouse-operated software synths wouldn't.

A small selection of hardware synths, by rough ascending price :

- The cheap and fun Korg Volca range mentioned in the article. - The Boutique range from Roland. Spot-on virtual analog (=DSP-based) reissues of their classics, the JU-06 is probably the easiest to figure out and most faithful to the original Juno 106 (a staple of the 80s), but it's sold out in most places. - The Arturia microBrute and miniBrute. They spearheaded the cheap analog renaissance a few years ago and sound lovely and gnarly - The beautifully designed Korg Minilogue mentioned in the article. - My all-time favourite interface on a synth is the Nord Lead 3's. It's full-featured but especially suited to beginners because of a truly unique feature : endless rotary knobs with LEDs indicating current value. It's virtual analog (DSP-based) with everything you'd expect on the substractive synthesis front, plus the easiest FM synthesis interface you'll find anywhere. It's been discontinued for years (and its successors don't have the LED knobs), but some units pop up on eBay regularly. - At the higher end of the price range, there are also dozens of options : the Prophet 6 and OB-6 by Dave Smith Instruments (the inventor of MIDI) are poised to become modern classics. The DSP-based Virus TI is incredibly versatile (although it's getting long in the tooth, as CPU power has caught up with DSPs meaning a plug-in like Serum is just as good)


I don’t have any recommendations for reading, but do enthusiastically recommend the korg volcas from the article - they are self contained, fun and intuitive to use, and very affordable. They sound good enough that you will continue to use them even if you have more expensive gear, and what you learn about synthesis transfers easily.


Honestly, Pick up a Volca Beats and a Volca Bass and start experimenting. After a few weeks of playing with both, you'll start having more questions.

If you buy them used, you should be able to sell them for what you bought them used for.


Tatsuya Takahashi understands the importance of accessibility in design. He has lowered the barrier to entry for new instrument buyers that want to get into analog synthesis, and that’s what has made him so successful.


Not bad $300 retail. Looks like a great way to get into microtuning and atonal scales for electronic music as well ;)

https://www.midi.org/articles/microtuning-and-alternative-in...


If you're interested, you should check out Roland's SE-02 https://www.roland.com/global/products/se-02/

Also, here's a nice video on 3-Oscillator synths: https://www.youtube.com/watch?v=O7RlLZrnugQ


i could be wrong but i owned a minilogue and don't think it supports microtuning. only some really fanxy synths do.


They recently released the Monologue, which supports it.


"[T]he stuff I was making was really weird, because I didn't have real-world, existing examples to work from. I mean, they were around, but I guess I wasn't interested enough. It just seemed beyond what a kid could buy with his pocket money."

His good intuition and willingness to disregard tradition seem to be the key to his success. Studying examples of the "right" way to build something might be a useful learning exercise, but it can also be mentally constraining.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: