View Full Version : Advice on recording engine noise for analysis

Nov 4, 2008, 04:54 PM
So, to start, digital audio recording is way out of my field of expertise. Photograph, yes, video, somewhat, computer science, booya.
But Audio? I'm just about clueless.

So here is my self imposed project:
Record the sound coming from my engine to help diagnose it.
The technical aspects I'll deal with, but in a "perfect world", I'm wondering about a few things.

What I'm looking for is advice in a couple areas:

1. Recording quality:
Can a simple line-in mic to a Powerbook G4 record the sound waves to detect 60 unique sounds per second? (a 6 cylinder at 600 rpm). How about 600 per second? :)
I am assuming that the engine is producing clear and distinct sounds that could be picked up.

2. Analyzing software:
What program on the free/cheap side would enable me to analyze the results, assuming I can get it recorded? I have GarageBand on my MacMini. While an automated analyzer would be nice, I am expecting that I was have to visually look at the wave form produced to pick out spikes in sound volume. Suggestions?

Any and all advice, comments, and suggestions are welcome!

Thanks everyone for reading. I'm aware this is a little out of the arena of the usual audio topic here. However, asking a siilar question on a car mechanic oriented form just gets blank stares :rolleyes:

Nov 4, 2008, 05:07 PM
so CD quality audio is at 44.1khz / 16 bit. So it is sampling the audio coming in 44100 times a second.
now it is very common practice to record at higher bit rates (24it) and higher sample rates (48khz, 96khz), and then bring it back down to 44.1/24 after the fact.

but in saying that, there's a lot going on that effects the quality of the recording, not just the sample and bit rate. and these things include:
Word Clock

for recording an engine for diagnostics, i wouldn't worry too much about most of that, and might recommend a USB mic. (it has the mic, preamp, and converter built in). (at this point dont worry about clocking).

This would be the simplest thing for you and it would provide decent enough results to do what you need. And i've never used garage band, but that should be a good starting place to get your samples recorded.

Nov 5, 2008, 11:28 AM
Thanks for the response, quite informative. I didn't realize, or never really thought about what the frequency rate translated to.

With your suggestion of a USB mouse, is this due to a normal external 1/8 jack mic being unable to really sample that fast? I know, I'm using a digital term to describe an analog process :)

My only concern with USB is:
Do I need "nice" software to be able to record in this way? I've never seen "from usb" as an option in most recording programs. Perhaps this is something Garageband allows (I'm not on my mini to check) ?

I have a question about analyzing sound. When looking at sound in a wave form, what is usually represented? Pitch? Are their programs that allow you to look at different parts? Certainly something like pitch might be useful (Injectors for example make a distinctly high "tick" sound), but volume is probably the better initial analyzing characteristic. I don't remember seeing a way in Garageband to analyze the volume output of a file on a milisecond basis.

Sorry, I know these questions are going beyond (or existing below) the normal procedure for musicians. I really appreciate the help I've gotten so far though, it gets me pointed int he right direction and gives me the hope I might get something out of this project, even if it's only a little education :eek:

Nov 5, 2008, 11:50 AM
i can't really help you analyze your stuff, i don't really do that. But wave forms are simple physics.

pitch = frequency ... the closer together the waves are (the more frequent they happen), the higher the pitch

volume (a bad word) = amplitude ... the taller the wave form, the louder the sound... but be careful when reading amplitude. it's measured in decibels, which are Logarithmic. so ever 3db the sound doubles in amplitude.. (eg. 123db is twice as loud as 120db)

i suggested this, not cause a normal crappy mic that has 1/8" out couldn't sample fast enough (that's not up to the mic), but cause a) that mic is bound to be horrible, and b) you're relying on your built in soundcard to do the conversion and preamp stuff, and usually those aren't good at all.
you would get a lot better recordings from something like this (http://www.sweetwater.com/c981--USB_Mics) (i would trust Blue, Rhode, and AT)

when you use that, you're going to have to install drivers for it, it will act like a soundcard, and your software will see it as an input.

Nov 6, 2008, 03:32 AM
The graph of a waveform represents amplitude over time (a.k.a. a time domain signal), so in terms of sound that represents variations in air pressure as the sound moves through the room or the amplitude of the vibration of your eardrum. You can kind of "see" pitch in that the closer together the peaks are, the higher the frequency/pitch. Unless you were to sit counting peaks/second (not a good use of time) you can visually read pitch from a waveform.

Before I give any analysis tips, what would actually be meaningful information for you to use to diagnose problems. Do you want to know if things (pitch/amplitude) are constant or irregular?

About miniplug vs. USB mics. As Drumjim says, a crappy 1/8" mic will sound bad, but by the same logic, a non-crappy 1/8" mic won't. Yeah, it isn't as good as using XLR or 1/4" and a good interface, but the connection is not necessarily a sign of quality of the mic. I use a Sony ECM-MS907 when I'm using my minidisc recorder in the field or using my laptop and I don't have my XLR or 1/4" input options available. For the price, I am impressed with this mic. It is a stereo condenser mic with either 90 or 180 recording angles. Honestly, for what you are doing, mic quality may not make a big difference (I may revise that when I hear more about what you want out of analysis). All that being said, if I recall correctly from before mine was fried, the PowerBook G4s only were line-in, not mic in, so my Sony mic would give WAY to low of a level to be useful--but it is worth double checking that, because my G4 is no longer alive and I know some Apple products are able to accept both line and mic level signals. Signal level is a much bigger issue in the 1/8" vs. USB debate than sound quality (but don't use adapters to get your mic down to 1/8", if you do and you are doing mono->stereo conversion with the adapters you can actually pick up radio frequency interference on the second channel).

P.S. Drummerjim, what does that sig quote mean? Maybe I'm just too tired to get it.

Nov 6, 2008, 11:55 AM
Thanks again for the replies.

As far as one I would like to look for in the results...
I'm hoping for consistent, and looking for irregular.
I will be looking for differences in rhythm I guess.

For example, the injectors make a sharp "tick" sound every time a cylinder fires. If I was able to identify that sound visually, I could run the engine and see if the injectors are consistently firing under a multitude of conditions.
I don't really care about little discrepancies between the sound that is made, as long as the sound is present and recognizable.

Does this sound possible?
I would know what the frequency of the peak signal is (because of engine RPM), I would just need to be able to recognize which peak amplitude went to what sound, yes?

newbie confession... I didn't know there was a differences between Line-In and Mic-In. Presumably one has greater preamp abilities to deal with a weaker signal?

Nov 6, 2008, 12:28 PM
To do what you want to do, you probably will want to look at frequency spectrum, as well as the timing.

You can get a decent freeware analyzer to play with here: