Pre-Purchase Question re: PMW-EX1

basspig

basspig

Veteran
This seems like a simple question, and pretty important, given the level of criticisms that the PMW-EX1 has received so far in several other areas, but one that no one has yet been willing to or able to answer, and the lack of information on this issue has been holding up my purchase of a couple of EX1s for at least two months now:

How well does the audio performance measure up?

I've come to realize that just because B&H lists the audio frequency response as "20-20,000Hz" on the specs page, does not mean it's so. (Tested my HVR-V1Us--both of them--and they are only flat from 1,222-16,000Hz). I'm hoping that the PMW-EX1 will actually have some bass response so that I can record decent audio without the hassle of dragging along separate audio gear and the hassles of synching it all together in post.

Sony claims it's 20-20,000Hz +/- 3dB, but I want to know if it's true in actual production units that people own, and where that -3dB dropoff is (at the high or the low end of the spectrum).

Anyone can easily test their camera, using a free version of RightMark Audio Analyzer, and a good sound card, or, a CD burner and a CD player (to make the test signal source that the camera will record from). The whole process takes about 5 minutes and yields a large amount of data about the camera's audio system, including dynamic range, distortion levels, etc.

This is really the last impediment to me signing off on a purchase order for two PMW-EX1s, so I'm really hoping someone who owns one is willing to test it and post the results.

Most of the video forums seem to forget about audio entirely, focusing on subtle vignetting, or rolling shutter issues, but I have not heard but a few extremely infrequent concerns about the sound capabilities. And after all, it IS a digital recorder, so why shouldn't it be as good as a $200 Zoom H2 or H4 at recording audio? Hopefully someone who things his money is well-spent with the EX1 also considers the audio important enough to care about and will comment.

"Sound is 70% of video." (not my quote)
 
Last edited:
Actually, sound is only about 60% of video, according to my own tests.

Where did you get that crazy 70% figure?
 
matthew77 said:
Actually, sound is only about 60% of video, according to my own tests.

Where did you get that crazy 70% figure?
Click to expand...

Probably was some movie director, years ago, but has been repeated by the admin of another video board that is pretty highly-regarded.
 
Richard,

I am not sure how you were able to measure it with such precision, but your number is certainly closer to mine and therefore supports my findings.
 
According to this article, about 60% of the way to the bottom of page:

http://www.ajr.org/Article.asp?id=4428

...it is 70% of video. ;)


All kidding aside, I DO wish to make a decision on the XDCam by the middle of February. Since no one has it for rent around here, I have to rely on the online community to answer some of these questions.
 
Last edited:
myCharlie said:
Okay, I'm very new to sound. What is that number means and how does it translate to quality?
Click to expand...

Frequency response is a very important number. Our own ears are said to respond from 20-20,000Hz. (In practical tests, I have found people can hear down to 10Hz quite well, if it's loud enough, and youngsters can hear up to 25,000Hz and slightly beyond).

An audio recording system, so as not to come up short on recreating the sound of a live environment (a cityscape, for instance) should be relatively flat (have equal sensitivity) over a range of 20-20,000Hz.

The measure of the uniformity of the response is the decibel. 3 decibels (dB) represents the smallest change that the untrained ear can reliably notice. 10 dB represents 'twice as loud' or 'half as loud' difference.

20Hz sounds like the sound of an army helocopter approaching the horizon. Window frames on old house rattle to this sound. If you have been to church, you've probably heard tones in that range from the pipe organ, if it is a large cathedral.

20,000Hz is very high up. Most people over 40 can't hear it. But all of us have owned a television set, and many can probably remember hearing the high pitched 15,700Hz whine from the television's horizontal sweep flyback transformer.

The upper frequencies are important because that's where the harmonics of common sounds exist--they give character and identity to the sounds.

The lower frequencies are important because that's where the fundamentals of music, and just about all sounds in an industrial world exist: the rumble of the subway, the sound of thunder, a door slam, a bass drum from a marching parade, etc. If these frequencies are missing, we say the sound is "tinny".

Modern digital recorders are typically flat (that is they maintain a constant sensitivity) from 20-20,000Hz plus or minus one-tenth of a decibel. ( +/- 0.1dB )

Modern digital video cameras record sound digitally. It stands to reason that if the manufacturer makes no effort to alter the normal frequency response of the A/D-D/A chipset, the response should be from DC to just about 1/2 the sampling or Nyquist frequency. Cameras typically sample at 48Khz (48,000Hz), so the upper theoretical limit of audio is about 24KHz, but in practice, due to protection from aliasing artifacts and filter limitations, it's usually cut off at 21KHz. On the other end of the spectrum, A/D converters are flat to DC, so 0Hz on up, basically.

So using off the shelf chips, any video camera that doesn't have additional circuitry to limit the response will be flat from DC to 1/2 the sampling frequency.
 
Well explained, basspig.

But thought that 1dB was the "just noticeable difference" (JND) and that 3dB was a doubling in power and hence twice as loud.

Regarding the test of the camera, I do not have a PC and cannot help you with the test using your recommended software, unless you can come up with a way for me to do it with my Mac.
 
That's true for experienced listeners and audio engineers, but tests on the general public have shown that 3dB is the threshold that the untrained ear can reliably detect. Personally, my old computerized EQ has 2dB steps and I find that to be too coarse--I need something in between, say, +4 and +6dB on an EQ band.

The scale is 3dB= doubling of power
6dB= quadrupling of power
10dB= ten times power
1dB= 25% change in power

Can you play a WAV file? Or better yet, burn it to a CD? Do you have access to a CD player that you can patch into your XDCam EX' inputs?

What I could do is generate the test signal (RightMark creates a wav file with a FFT test signal, you could burn that to a CD and cable the output of the CD player to the camera, setting the inputs to line level, or using an attenuator to match the line level out from the CD player to the Mic level of the XDCam.

Record the test signal on the XDCam, and export the clip. Perhaps extract the audio portion from your NLE to a wav or other PCM file at native bit depth and sample rate of the camera. Send that file back to me, and I'll feed it to RightMark for analysis (it compares the original FFT test file to the audio recording of that test signal and derives about a dozen different audio characteristics from that information. I could then send you a zip file with the data.
 
Last edited:
Sounds doable - I can get a wav or aiff file into the the EX1.

I might have some issues getting it in a mic level to test the mic preamps - my 1604 VLZ & computer will be way too noisy if I try to set the output that low, so perhaps I'll just go in at line level.
 
Two possibilities for getting the audio into the EX-1:

1. a high quality Firewire connected sound module, like M-Audio, MotU, etc.
2. Burn the signal to a CD and use a CD player's line outputs.

Two ways to test:

1. Line level (that's what I ended up doing with my V1U)
2. Mic level (you'll need an inline attenuator, as you'll be amplfying noise if you simply take the line out of a mixer with the faders potted way down.)

In line attenuators are pretty common and in most event videographer's toolkit, especially if you're in the habit of accepting live mixes off FoH mixers, so you may already have some.

Not sure how large an attachment I can put on here, but if not, I can send a YouSendIt.com link and you can download it that way.
 
I do not have any passive attenuators, so it will have to be line level. I'll be going from the sound out of my Mac, or perhaps my firewire deck (sony DSR-11) which may sound better.
 
I know that I have got to be missing something in this thread, maybe I should go back and re-read it. But if you are playing out of a CD player and into a camera, the audio quality will only be as good as the A/D converters out of your CD player which can vary dramatically between brands. If you have to go into a Mackie or something first, you are degrading the sound even more no matter how small it may be. Whatever you end up with in the end, it will only be a rough estimate. Am I missing something? Hmmmmm.... it is 3:20 in the morning, that could be it.
 
matthew77 said:
I'll give it a go tomorrow. What level should I record at?
Click to expand...

Try for just under clipping, about -1dB, or, lighting up the second to last dot on the meter scale (but not the red one that indicates clipping).

There must be NO clipping at any point in the recording, else RM will yield an error message instead of results.

Important note: Be sure to disable any wind filter or other sound filtering, noise reduction, etc that might be available on the camera, as these options may have drastic effects on the response curve of the camera.
 
Last edited:
TheMusician said:
I know that I have got to be missing something in this thread, maybe I should go back and re-read it. But if you are playing out of a CD player and into a camera, the audio quality will only be as good as the A/D converters out of your CD player which can vary dramatically between brands. If you have to go into a Mackie or something first, you are degrading the sound even more no matter how small it may be. Whatever you end up with in the end, it will only be a rough estimate. Am I missing something? Hmmmmm.... it is 3:20 in the morning, that could be it.
Click to expand...


Generally correct, for critical measurements, however, in the case of the cameras, the D/A converter in your average CD player is going to be much superior to that found in a typical camera. Anyway, my intent is to get a rough idea of the response curve on the XDCam, to find out if there's a high pass filter or not.
 
You must log in or register to reply here.
Back
Top