Audio Analyzers: Pink Noise vs Sine Sweep with Michael Lawrence from Rational Acoustics
Vložit
- čas přidán 22. 07. 2024
- Why do some audio analyzers default to pink noise and others to a sine sweep?
They are different acquisition methods, but they'll both get you there.
See the full post: www.sounddesignlive.com/audio...
Real-time
- 2 channel input
- Measure in real time with any source material
- Common among live sound engineers
Non-real-time
- 1 channel input
- Measure off line with a pre-determined source
- Measure THD
- Get very high SNR
- Characterize decay (T60, C10)
- Common among acousticians, manufacturers, mobile electronics technicians
You don't want to feed a real-time measurement with a sine sweep because the real-time analyzer wants a broad-band source, which makes the sine sweep look like mostly nothing. It's like putting unleaded in a diesel engine.
___
📈 Get Started with Sound System Tuning - www.sounddesignlive.com/get-s...
___
👨🎓 SOUND SYSTEM TUNING ONLINE COURSE - www.proaudioworkshopseeingsou...
___
📕 Books on sound system tuning - www.sounddesignlive.com/audio...
🎙 Podcast for live sound engineers - www.sounddesignlive.com/pro-a...
___
Smaart® and the Smaart logo are registered trademarks of Rational Acoustics LLC and are not affiliated with Nathan Lively or Sound Design Live.
___
Be friendly
Facebook - / sounddesignlive
Twitter - / nathandofrango
LinkedIn - / nathanlively
___
I love to geek out about the physics of sound. This channel focuses on the growing opportunity for live sound engineers to improve their confidence and consistency through the understanding of the principles of sound system design and optimization. My goal is to make this channel upfront and honest about my success and failure, so you can learn from both.
I am always open to suggestions and feedback so please comment on this video or contact me through my site. - Věda a technologie
Have you tried both real-time and non-real-time measurements? Do you find one works better for you for a specific job?
Nathan, I just wanted to say thank you so very much for doing what you do and sharing all these wonderful and very informative videos, I know it's appreciated by many! I really like your style and please keep up the awesome work! Oh yeah, I too did the gas in a diesel truck thing once (hopefully never again) it ended costing me a little over a grand, still to this day I cannot pull up to a pump that has both diesel and gas without trembling from the traumatic experience and triple checking myself while filling up. Thanks again!
Thanks for having me, Nathan. I enjoyed the discussion!
Any time!
Thanks for commenting. I was on the hunt for your channel. Not sure why it wasn’t in the description.
Excellent video - thank you Nathan!!
Such an amazing and informative video. Many thanks! 🎉
Thanks for this video!
This was awesome. Thanks!
You're very welcome!
Thanks for the video. I have heard some people use short bursts of noise at different freqs, and/or clicks in systems without delay speakers. Is this for RT60 decay measurements at various frequencies, or something else?
Unsure! Maybe Michael knows?
From a measurement perspective, we need to pass a frequency through the system in order to get data at that frequency. In other words, if you try to use a flute recording to measure your subwoofers, that won't work very well. So for general measurement purposes, a broadband signal of some sort is important.
That being said, there are a variety of specialized test signals that are used for specific purposes, for example the STIPA signal used for speech intelligibility tests, the M-Noise signal for linearity tests, and so on. Some people use clicks or pops to try to set delay speaker timing by ear. RT measurements can be gathered with a direct impulse (pop/bang/clap/shot) although there are some drawbacks with doing so, and a sweep or other period-matched signal is generally a better approach.
All of these signals have certain mathematical characteristics that are designed to quantify particular properties based on the test that's being run. So in a sense it comes back to what Nathan and I discussed in this video: starting with a goal about what you'd like to measure, and choosing the appropriate tools to get you there.
-m
@@precisionaudioservices1181 Thank you, I appreciate your reply!
I've been figuring out how to calibrate my vinyl playback system. Sine sweeps seem to work out nicely out of the box, but for pink noise the lower frequencies get lower measurements (using the analyzer on the RME device - which doesn't use FFT). My interpretation would be, that due to the mechanical nature of the sound reading, there would be some inertia involved for the low frequencies, which doesn't get measured with a frequency sweep, but the pink noise is changing its frequencies so fast that the inertia has a measurable effect.
What would be your take on this? Are you aware of other examples, where sine sweeps produce a different result from pink noise?
Hey BoooM, I'm not sure, but it may only be a question of getting the level high enough above the noise floor. the sweep is automatically higher above the noise floor because its crest factor is lower, but either should work fine in low noise environments.
@@nathanlively Thanks, now learned to include the crest factor into my thinking process. But I don't think it's related to the noise floor. I just contacted the manufacturer of my phono preamp, hopefully they can help me out. I probably need another cartridge which synergizes better with my amp. My assumption is that the amp kinda amplifies the tendency of the cartridge to overemphasize the higher frequencies.
When testing the performance and allowable input of speaker units, do they usually define pink noise and test with sine waves without using pink noise, will the voice coil and unit be destroyed by the sudden temperature rise of the voice coil?
You may want to check out m-noise: m-noise.org/
Hi! Nice discussion!
But let's talk about something different.
I compared two methods: pure Dirac impulse recorded back to DAW, and regular sweep in REW. Nothing except mic gain and speaker volume was not changed. But I got slightly different results. My thoughts what it is much easier for loudspeaker to play sweep tones instead of Dirac, so i think for fast sounds Dirac is more accurate in terms of possibilities of loudspeaker, and for more harmonic signal sweep tone is more accurate. What do you think about it?
hey Michael, I haven't used Dirac impulse so I can't comment. You might reach out the creator of REW for his opinion.
@@nathanlively I made another measurement with Dirac and it was similar to sweep, so i think something gone wrong at first time.
I personally use IR when gathering info and then use the real time measurement when doing correction..
Cool! What does gathering information mean? Like a site survey?
I'm finding this video very late but anyway. My question is this:
Was the pseudo-random noise just a loop of some random noise? Or was it one of the exact and completely deterministic sequences (called "maximum length sequences") which sound like pseudo-random noise and which have been explicitly designed for the goal of impulse response recovery? It would definitely be nice if these had been discussed there as well.
Hi Petr. From the Smaart User Manual:
Technically, the random number generator is pseudorandom also, however it is randomly reseeded every time you start the generator and given its cycle length of 219937 samples, it will effectively never repeat.
Pseudorandom noise signals in Smaart
repeat on intervals that are a power-of-two
samples in length up to 219 (512K samples).
These can be band-limited or shaped to an
idealized long-term average speech
spectrum (LTASS) in addition to broadband
pink noise. When using pseudorandom
noise, you should always select a cycle
length that is at least as long as the longest
FFT size that you are using for measure-
ments. The longest time window used in the
MTW transfer function is a little over one second, so 64K would be the lower limit in that case but noise sequences even that short tend to become hard to listen to pretty quickly.
Thank you. From the description it seems that these are actually not MLSes. But anyway, the most obvious way to know if they're MLSes or not is that an MLS only alternates between two and never more than two different sample values (usually +1 and -1 or +1/2 and -1/2). Also, one important property of MLSes is that if one sound file contains the same sequence repeated over and over and another sound file contains the same sequence played just once and backwards, convolving the two sound files makes a single impulse repeated, provided that the sample rates of the two sound files are identical and that no intermediate resampling has been done on any of them. Nowadays MLSes are no longer popular because they often give a worse SNR than what you get with a sine sweep and because it's often not possible to guarantee the same sample rate both for the recording device and for the playback device.