Dave Hill | TapeOp

David Hill is the owner and designer behind the company, as well as the owner of a recording studio with a lot of customized components, not surprisingly. Compressors, preamps, digital converters, tape machine components — David Hill has had a hand in designing gear that is fast becoming legendary in the pro audio industry.

Where did you get the name Crane Song?

It originates from when I was working on industrial film, a documentary.

How'd you get started in gear design?

I started when I was very young. I think it was around first grade or even younger. A kid that lived next door was messing with radios so I started messing with them as well. I messed around with radios, analog synthesizers, and then I started working in local music stores fixing everything — Moogs, Wurlitzers, Marshalls, Fender amps. Some friends started a studio, so I got involved with that fixing things at first. They had bought an Ampex MM1000, a 1" 8-track machine, and asked me if I wanted to become an owner to help pay for it. Eventually I ended up owning the whole studio. I also did Summit Audio, doing designs up until 1994.

Are you self-taught or academically trained?

Self-taught. I did live sound for a rock band, went on tour to Florida, came back and enrolled in a tech school for an electronics degree, a piece of paper.One spring, before I even graduated, one of the instructors left and I applied for the job. I taught there for eight — and-a-half years. My involvement in Summit ended and a year-and-a-half later my first product, a compressor, came out in 1995-1996.

What kind of converters do you use in your A/D converters?

Crystal Semiconductor parts, which I feel are more musical sounding.

And you create the analog stage and interface?

Yes, as well as the DSP [digital signal processing] and the components that help bridge the analog and digital worlds. All the pieces I design are constantly being used/refined in my studio. Gear design is a big use for my studio. Designing a piece of equipment is a bit like doing a mix. You get it basically up and working and then tweak it to make it sound like you really want it to. I've been known to change out parts in the middle of a session.

It sounds like your studio is a combination laboratory and QA station.

Yes, there are definite advantages for a gear designer to have a recording studio. I designed the control room myself. I did a lot of research and I've had the advantage of being in a booth next to people from Walters-Storyk at AES, and they've been very helpful. It's the third control room I've built and it's a really good-sounding control room. Things translate very well. It was also years of work to get it so everything did work well.

What kind of recording projects do you tend to get?

Right before AES I was mastering some rock 'n' roll that was recorded here. I just finished working on a project with a Finnish musician. Next week I've got a Norwegian thing that's got to be done soon for a grant. I recently did a project with a producer, Scott Matthews, and an engineer, Tom Lukens, from Mill Valley [SF Bay Area] where they flew out and stayed here for the duration of the project. It was for a band called Almost 7, which I played second engineer on and mastered. I've got three choir projects that I have to record, edit, master and be ready to duplicate soon. So I'm tending to be on the busy side.

Okay, here's a digital question: What is dither and why do we need it?

Dither is a noise, like pink noise. It's a big enough deal now where gear designers can buy someone else's dither for a lot of money. Dither randomly moves around the least significant bits, so instead of being just truncated, or cut off, (example: when you go from 24 to 16 bits), it provides randomness there. It depends on the material, and how it relates to the sampling frequencies. If your frequencies are submultiples of the sampling frequency, or close to that, the dither is a lot more important. When audio is truncated, you get distortion, which dither helps to randomize and minimize. One of the things about Nyquist's sampling theory is that you can reproduce audio, but it does not say reproduce without creating distortion.

So what's the deal with your Crane Song "Analog Generated Dither" CD?

It is an analog-generated noise dither source. You get the CD and dump it down on to your pair of stereo tracks. Knock it down 25 dB, and mix it in. Leaning to hear what dither does takes some time, you have to learn what to listen for.

Do you do this at the 24-bit or 16-bit stage?

Sometimes while converting, sometimes not. Here's the thing: you get a different result if you dither before truncation versus if you wait until after truncation. It can affect EQ, so it's one consideration to add dither before or after you EQ. I try to do it when doing EQ or level changes. When going from 24- to 16-bit you must dither or your audio quality will suffer.

Now that we're getting deeply entrenched in the "digital age" of audio production, would you say that we've reached the pinnacle of pro audio gear design and overall knowledge when it comes to analog-based gear?

There are always things that can be learned and things that can be done better. Analog will be around as a niche thing at least for a while. It will never fully go away and it will get better yet. The reality is that analog does sound better. Anyone will tell you that.

Yeah, I hear it all the time...

Digital systems have some amazing editing tools. You can do things that you could never do on tape. But it's important to use it in a way that keeps the energy and life of the music. One thing about tape is that you have a thing called head bump. The frequency response is never really flat, it moves around, up and down, up to a dB, which can add low end and change other things, but is not as accurate as digital. Low quality gear, both digital and analog, sounds bad.

What do you think of the 96 kHz sample rate as becoming the standard, at least within pro audio production?

96 k sounds better, especially in the signal processing stage. EQ sounds better. I think that 96 k is just starting to become workable at this point for most of the world, just because it's finally evolved to the point where the tools exist and they're becoming affordable. I think that 96 k will eventually become the standard as far as multi-track recording is concerned. There are still a lot of people doing 48 k stuff, and it will take awhile, but I think it will happen.

Have you thought of building a lower- priced piece of outboard gear or converter box?

Here's the problem: I've gotta be happy with it, or I'm not gonna do it. Part of the problem is that I use conductive plastic potentiometers, which cost almost five dollars each. Toggle switches are six dollars. Rotary switches are eight to ten dollars. Front panels end up at $35-$40. Power supply and chassis cost us $120 by the time it's done. You are at $300 without audio circuits and labor. You aren't gonna put a lot of audio in there and make it sell for a grand. The discrete stuff does sound better, but it costs more.

Speaking of discrete, what about your new EQ?

The Ibis has four bands with a low cut filter and musical note centers (as opposed to frequencies), and yes, it is Class A discrete. As is the case with most Crane Song products, it has a color generator that allows [you] to change the color of any one of the four bands, or the program — by adding second and third order harmonics. The response to the EQ at AES was extremely enthusiastic.

What about the Avocet?

I'm building the Avocet mostly as a monitoring device for DAW stations. All of the digital signals are up-sampled and jitter-reduced. It's meant to be sort of the center module of a console in a DAW environment.

What about your quest for decent 192 k converters for the HEDD?

I am working on the HEDD 192 at the moment. It's a little bit more difficult than just swapping out chips... there are all sorts of issues involved — you know, making sure the DSP is going to be fast enough, not compromising anything that was going on. There are all sorts of things involved that have to be dealt with, so it's not exactly an easy task. It's really just going through and optimizing what I'm doing... and then there's some tricky things. What happened in converter evolution is that when you went from the 48 k parts to the 96 k parts, it basically just doubled all of the clock rates. So that's easy, you just scale everything that way. But when they went to the 192 k rate not all clocks doubled. The master clock that everything runs off [is] still at the same rate as it does on the 96 k part, and as a result of that things don't quite scale the way you think they do, so you've got to do some tricky stuff.

What is "up sampling" and how does it work within your and other people's designs?

The idea of up sampling is to change the sample rate to a higher rate, as an example, change a source sample rate of 44.1 k to 96 k. If it is done very well it can sound good. If it is not done well it will sound like crap. One of the uses of this is to do signal processing at the higher rate, because there will be less nasty digital artifacts happening to the audio source when processing at the higher sample rate. The result is better sound if it is done well. Another use is to create an up-sampled D/A converter. Doing this has the potential to gain audio quality in several ways. The first is that when using a sample rate converter that is locked to very good crystal on its output side, there will be jitter reduction by the SR converter before the digital to analog conversion. In theory there will also be benefits as a result of better reconstruction filters during the conversion. The filters will not be as steep. This use of a sample rate converter can also be used on jittery sources, for jitter reduction only with its output being digital. It can even have an output same sample rate that is the same as the input sample rate, [but] the output clock source must be very stable. Jitter reduction can help a lot on questionable sources. However, nothing is free — sample rate conversion does add latency.

What do you think of the way a 192 k recording sounds? Have you engineered any of your sessions at 192?

I have experimented with other manufactures' hardware and I am not that impressed with what I have heard so far. I do get a few requests. But because of the DSP involved, I don't think you're going to see any rock 'n' roll being done at 192. I think you're going to see a little bit of classical and jazz and a little bit of mastering stuff.

So now you're also doing a plug-in?

Yeah, the Phoenix is a TDM plug-in with different flavors of tape color — similar to the Tape Knob on the HEDD. The Phoenix integrates a very efficient use of the DSP on the cards — it's multi- shell, so it's very efficient. It's Pro Tools HD, Accel and Mix only.

It was great to talk to you. Anything else you want to add?

Designing gear is sort of like building a mix, a lot of the time it's about compromises. You do your best to make it work the best. In a mix hopefully you have good tracks so you don't have to hide the ugly stuff. Hardware is kind of the same thing. I figure out what I want it to do, I get a picture in my mind [of] what I want it to do, and then I build some circuits and start listening, and then I ask myself, "Does this really work?" That's really how the gear evolves.