User endolith - Social Sound Design

Answer by endolith for Workflow Question: Batch splitting stereo tracks into mono?

2010-12-12T15:33:33Z

I'm sure SoX can do this, but you'll have to figure out the right command.

http://billposer.org/Linguistics/Computation/SoxTutorial.html says:

For example, to extract the left channel give a command like this:
  sox foo.wav -c 1 foomono.wav avg -l

Answer by endolith for Radiofrequency hearing

2010-10-31T04:21:24Z

http://en.wikipedia.org/wiki/Microwave_auditory_effect

I'm sure you're not hearing RF from your LCD. Record it with a microphone and look on a spectrum analyzer, and you'll just see some tones in the kHz from the backlight electronics.

Answer by endolith for Polymer Film Microphone

2010-08-17T00:09:42Z

Piezoelectric pickups need a preamp with a really high input impedance, preferably on the pickup end of the cable rather than the amp end.

Piezos have a very high (capacitive) output impedance, so they need to see an even higher impedance at the input of your preamp, or they'll be prone to interference and low-frequency roll-off.

(A 500 pF pickup, for instance, has a 6 MΩ output impedance at 50 Hz, so even if you plug it into a "high-impedance" 1 MΩ guitar input, it will have dropped off by -17 dB at 50 Hz.)

http://liutaiomottola.com/electronics/bassducer.htm

Answer by endolith for What do colors sound like?

2010-08-21T20:10:43Z

Yes, you can convert color to sound.

No, orange doesn't have a particular frequency or sound. You can pick whatever mapping you want. Do you want orange be a specific sine frequency, or should it represent a square wave? You pick.

AudioPaint and Coagula use the color of a pixel to determine the panning of the sound.

Spectrograms often use color to represent the amplitude of a frequency component. Software like Adobe Audition will do the inverse function to convert an image back into a sound.

I've been trying to go the opposite direction, using audio frequency content to determine color of a waveform:

This is the same method used by Freesound.org, using the spectral centroid of the sound to produce a monochromatic color, but shaded to try to show the density of the samples as well as the peak amplitude.

What I'd really like to do is map the spectrum of the sound onto the spectrum of the color, so a low frequency sine would be red, a high frequency sine would be blue, and a combination of both would be magenta. Last time I messed with it, I was trying to figure out a way to weight it so that logarithmic sine sweeps look like a rainbow, white noise is white, and pink noise is pink. I'm not sure if it's possible, though, because of the conflict between linear and logarithmic frequency axes.

Answer by endolith for High Altitude Impulse Response

2010-09-12T05:05:21Z

2:46

Answer by endolith for What do sensitivity measurements of microphones really mean?

2010-09-02T20:51:48Z

Sensitivity measures the voltage at the output pins of the mic for a given sound pressure level. 13 mV/Pa means that an air pressure oscillation with an amplitude of 1 pascal (94 dB SPL) at the mic capsule would produce a voltage oscillation with an amplitude of 13 mV at the XLR pins.

Gain is basically the ratio of output level to input level.

In this case we're not talking about an input and output, but you can still talk about the relative gain difference, which would be (25 mV)/(13 mV). The sound pressure level (1 pascal) is the same for both, so it cancels out. The relative gain difference is then 25/13 = 1.9, or about 2x as much voltage for the same pressure.

In decibels, the difference would be 20·log₁₀(V₁/V₂), so +5.6 dB for your example. A doubling in voltage is about 6 dB, so you could guess this just by recognizing that 25 is almost twice 13.

Answer by endolith for How loud is 0dB, and what is a good dB to mix to?

2010-08-23T20:58:26Z

"0 dB" is unity gain. dB by itself is a relative measurement, not an absolute level. It just refers to a gain unless you give a reference.

Perhaps you're referring to "0 dBFS"? This is the absolute maximum a digital system can measure (clipping), and in this case, your signals should only reach 0 at rare peaks, if ever. The average RMS level should be significantly lower.

Or maybe you mean "0 VU"? This is equivalent to an electrical level of 0 dBu, which is typically about 25 dB below clipping. In this case, the level should generally hover around 0 on the meter.

The right level depends on the crest factor of your signal (Wikipedia says 12–18 dB for a processed mix or 18–20 dB for unprocessed recording), and the dynamic range of the system. If you record too high, you get clipping. If you record too low, you get noise. You need to find the best point between the two.

Answer by endolith for What's your strategy behind 192kHz recording?

2010-08-21T20:34:38Z

There's no point in recording frequencies above 20 kHz unless you're going to do something with them, like put them through an effect that will modulate or stretch them back into the audible spectrum.

The main benefit of higher sampling rates is not to capture ultrasound, but to reduce audible things like aliasing, phase distortion, etc.

These can be a benefit for synthesis, too. If your square wave generator is poorly-written, it will produce an infinite number of harmonics, which are aliased and sound bad. If the sampling frequency is higher, this effect is reduced. Of course it would be better if the generator were written correctly, but you can't always control that.

Answer by endolith for Who uses Freesound.org on a regular basis?

2010-08-20T21:30:40Z

Lots of free (and terrible) recordings.

There really needs to be some kind of quality control, rating system, etc. that actually works.

Answer by endolith for So, what do we do when we don't want to ask questions?

2010-08-17T16:16:11Z

The whole point of Stack Exchange is to prevent chatty discussions. It exists because "just talking" has a terrible signal-to-noise ratio and is a waste of time for anyone who's just trying to find an answer to something.

There are already tons of websites for socializing, called forums. If you end up on one from a Google search, you have to wade through dozens of pages of people arguing, stating erroneous things as truth, stupid signatures, etc. in order to find a tiny partial nugget of information which may or may not be outdated or wrong.

http://stackoverflow.com/about

http://meta.stackoverflow.com/questions/38315/partner-with-traditional-forums-to-support-open-ended-discussions#answer-38323

http://meta.stackoverflow.com/questions/8658/do-you-feel-stackoverflow-needs-a-general-discussion-forum

Answer by endolith for Should I record 24bit 96khz and then dither down for distribution?

2010-08-14T03:44:45Z

Always record at the highest bit depth you can, so you have the best dynamic range to work with. After you've mixed and gotten the levels right, then you can convert to 16-bit.

Dithering is used when converting to a lower bit depth. It decreases nasty things like distortion by increasing the overall noise a little bit.

As for sampling frequency, I'd record at 88.2. It's not a matter of recording what humans can hear; 44.1 is plenty for that. Instead, it's a matter of avoiding aliasing. All converters have some amount of aliasing. Recording at a higher sampling rate ensures that the aliasing stays way up in the ultrasound, and then you can get rid of the ultrasound with a really good digital filter on the computer instead of the reasonably good digital filter in the oversampling ADC.

And 88.2 instead of 96 because 88.2 is a perfect multiple of 44.1. To downsample, you just do the digital filter and then drop half the samples. To convert 96 to 44.1, you need to, uh... do more than that. I'm sure they don't literally oversample by 147x and then downsample by 320x, but that's effectively what the algorithm is doing. It's more complicated than 88.2, so the processing will take longer, even if the converter is designed well enough to get the same output quality, so I see no benefit to 96.

Answer by endolith for An alternative to a blimp? (...a small pistol grip for example?)

2010-08-01T00:38:19Z

This could be less conspicuous than a blimp?

http://vimeo.com/4387137 http://www.redheadwindscreens.com/products/

Answer by endolith for re: recording ultrasonic frequencies

2010-07-29T19:15:14Z

Am I correct in assuming that the biggest hardware obstacle to recording sounds above 20k is generally just the microphone (not the pre or D/A)?

No, it's the hardware between the mic and the ADC, or the ADC's internal digital filtering, whichever cuts off first.

I believe most ADCs use digital filters that scale with the sampling frequency (the sigma-delta ADC datasheet I'm looking at does, at least). So recording at 96 kHz will let the ADC go up to ~48 kHz, for instance. But if the hardware before the ADC is rolling off before this, then those frequencies won't get to the ADC to be sampled, and you'd have to modify the analog hardware to raise the cut-off frequency.

The 744t specifies "10 Hz–40 kHz, +0.1, −0.5 dB", so it will let you go to 40 kHz. :)

Answer by endolith for Noise reduction on my library sounds? Yes or no?

2010-07-18T19:39:27Z

Always keep the raw files. Hard drives are cheap, and you can go back and do it better if you learn better techniques in the future.

Answer by endolith for Using a Bow for Sound Effects

2010-07-16T19:39:58Z

I'm a violinist, and I don't think you're going to hear much difference. The bowing technique and type of rosin will have more of an effect than the wood or hair, certainly.

You are using rosin, right? :)

Answer by endolith for in reality, do you believe convolution compression can work?

2010-07-02T02:38:39Z

Are you talking about modeling a real device by getting an impulse response from it and then convolving the impulse response with the signal to simulate the compressor? Using convolution that way only works for linear time-invariant systems, and compression is non-linear, so that wouldn't work.

Answer by endolith for MetaSynth 4 Pro

2010-06-28T17:46:02Z

Adobe Audition 3 does the sound-to-image-to-sound thing, and lets you do effects in the time-frequency plane using the lasso tool.

Answer by endolith for When will films be mixed at 96K?

2010-06-22T15:30:51Z

96 kHz and 32-bit floating point aren't for the final mix. They're for intermediate processing. In the final mix you shouldn't hear a difference between 48 kHz and 96 kHz (unless you're Batman or the mix was downsampled poorly). Likewise, 20-bit is adequate to represent all of the 120 dB dynamic range you'll ever get out of a real system (blame physics).

BUT, when putting a waveform through a lot of processing in a DAW, you want lots of headroom in both frequency and amplitude, to prevent unnecessary computation noise, aliasing, etc. Like photocopying a photocopy, the better the quality of each copy, the less degradation there will be in the final product.

Answer by endolith for What would sound sound like at the speed of sound?

2010-06-01T00:22:48Z

If there's no wind, then the air must be moving at the speed of sound alongside you, which would mean that you would hear in it normally, like the air carried in the cockpit of a supersonic plane.

Answer by endolith for Understanding Convolution Reverb

2010-04-13T19:02:45Z

"I was told to use a sound file that is a click followed by a sine sweep, followed by another click."

Hmmm... Who told you that? If you record the response to a click, then that is the impulse response. It's literally the response of a room to an impulsive noise, like a click. No deconvolution or other processing necessary. You then convolve your music with the impulse response, and it will sound as if you had played the music in the space instead of the click.

You probably want to use a dedicated clicker, though, rather than a speaker. You want the sound as close to an ideal impulse as possible, not filtered through a speaker's frequency and phase response. You want it to sound as if your performer is actually in the cathedral, not like you're playing a boombox of their CD inside a cathedral. You also want it as loud as possible (without clipping), so that the signal-to-noise ratio of your recording is high.

You can also derive the impulse response from a sine sweep or maximum-length sequence or other signal by deconvolving first. This improves the signal to noise ratio, but (ideally) it's going to produce the exact same thing as the straight impulse response. Practically, one method might produce better results than the other. See Wikipedia

See here and here for the use of maximum-length sequences to measure impulse responses. These work like a chirp, but better.

And remember you can record from two microphones at once to get a stereo image of the response.

Answer by endolith for How to buy music so that the greatest proportion of the money goes to the artist?

2010-04-19T14:15:05Z

Here's an infographic on the proportion of money that goes to the artist and label for different types of music sales:

How to buy music so that the greatest proportion of the money goes to the artist?

2010-03-09T03:05:25Z

(Copied from BPM StackExchange)

I assume that buying CDs at shows or from the artist's own record label website directly would be the best way. True, or not?

What's the second best way?

For instance, how much money do artists get when you buy mp3s on iTunes or Amazon.com?

Is it true that musicians make most of their money from selling merchandise?

Answer by endolith for DIY contact microphone

2010-04-14T19:13:48Z

It's not a matter of gain, it's a matter of impedance. Piezos have a very high (capacitive) output impedance, so they need to see an even higher impedance at the input of your preamp, or they'll be prone to interference and low-frequency roll-off.

(A 500 pF pickup, for instance, has a 6 MΩ output impedance at 50 Hz, so even if you plug it into a high-impedance 1 MΩ guitar input, it will be attenuated by -17 dB.)

Just an FET-input buffer will do. It shouldn't need any gain. Piezos can produce very high voltages on their own.

Answer by endolith for More reliable hard drives?

2010-04-15T17:49:56Z

None. I've owned Seagate, Western Digital, and Hitachi, and had probably half of each brand fail prematurely. They're all crap, and even the good ones will always fail eventually. Focus on having multiple copies of your data, not on a particular brand.

Warranty time is one way to choose. I think Seagate's is longer. I've sent in both Seagate and WD drives for replacement after they failed young. It's a hassle, but it's better than buying a new drive.

Answer by endolith for How to clean a digitally-distorted recording?

2010-04-11T17:04:17Z

Cool Edit/Adobe Audition has a Clip Restoration tool.

Answer by endolith for What are your favorite spectral tools?

2010-04-01T14:24:45Z

Adobe Audition (commercial) - Spectral Frequency Display produces a spectrogram for existing waveforms and Frequency Analysis can show a real-time spectrum
Sonic Visualiser (open source) - spectrum or spectrogram for pre-existing waveforms
Visual Analyser (freeware) - can show a realtime spectrum, too, though I don't trust the THD measurement.

As an engineer, I use them to see distortion, glitches or errors in sinusoidal waveforms (which may not be obvious on a waveform view, but show up as a bright vertical line in a spectrogram), aliasing in ADCs, estimate frequency, etc.

The latest version of Audition is really powerful, letting you draw lassos around parts of a spectrogram and then applying filters, gain, etc to only that portion of the time-frequency space. You can even import and export as image files, so that you can edit your recordings in Photoshop. :)

Answer by endolith for As a sound designer, do you buy 'audiophile' equipment for home use?

2010-03-20T14:09:55Z

If I didn't have such an overactive conscience, I'd definitely be in the audiophile business instead of the pro audio business. Ripping people off seems far more lucrative. ;)

Answer by endolith for 1-bit recording?

2010-03-20T13:50:04Z

Your 24-bit 48kHz uses the same oversampling sigma-delta modulation internally. Korg provides "AudioGate" software to convert 1-bit recordings to normal formats, which is what the 24-bit/96 kHz converter does internally anyway, so I don't see the point.

Answer by endolith for Legalities of sport TV in background?

2010-03-09T03:07:26Z

It's probably illegal, but, then again, so is everything else.

If there's no likelihood of being sued, you should use it illegally to protest against the ever-encroaching tentacles of copyright law.

Comment by endolith

2010-12-07T17:34:45Z

en.wikipedia.org/wiki/Missing_fundamental

Comment by endolith

2010-11-30T20:11:09Z

There will always be a small amount of error introduced with the conversion. Presumably the software sample rate conversion you are using has negligible artifacts.

Comment by endolith

2010-10-31T04:19:57Z

Johnson-Nyquist noise is the hiss you hear when you turn up an amplifier all the way. The thing we're always trying to get as low as possible in recording equipment.

Comment by endolith

2010-10-11T14:56:44Z

Where did you get them? Can you provide model numbers or links?

Comment by endolith

2010-10-04T15:35:49Z

Using copyrighted work for personal research is fair use, no?

Comment by endolith

2010-09-27T20:49:38Z

Using the same cables as a great audio engineer won't make you a great audio engineer.

Comment by endolith

2010-09-27T20:45:02Z

"I can send you my recordings if you'd like. There is a definite quality difference..." Multiple recordings of the same sound with identical mics in identical position through different cables, and with a definite quality difference? I'd like to see those.

Comment by endolith

2010-09-27T20:40:47Z

@Ryan, you trust salesmen way too much.

Comment by endolith

2010-09-25T14:56:06Z

Comb filtering? Are your mic cables several miles long? o_O

Comment by endolith

2010-09-25T04:16:29Z

The cable's capacitance could affect the sound if it rolled off the high end. Shouldn't happen with well-designed equipment, but with "high-end" tube equipment, you can never be sure... Also, make sure the cable doesn't generate its own noise from triboelectricity/electret/microphonics when you move it around. But usually the "quality" of a cable refers to the durability, not the sound.

Comment by endolith

2010-09-24T23:39:37Z

o_O What did you have it plugged into?

Comment by endolith

2010-09-24T21:50:02Z

Was it a blind comparison?

Comment by endolith

2010-09-21T19:15:15Z

Shouldn't this be a comment on the original post?

Comment by endolith

2010-09-16T01:44:57Z

Do what? This should be a comment, not an answer.

Comment by endolith

2010-09-16T01:44:10Z

It does work, kinda :) endolith.com/wordpress/2010/09/15/…