During capture I'm not sure how to do it, unless you're using a separate sound card that displays the waveform and can zoom out enough. Just seeing the audio levels won't do it. I've never done it that way, so I'm not sure if it would work. After capturing any program that displays the waveform you'll see the flat tops like in the screen shot in the first post of this thread. Using Virtualdub, just load the captured file and right click in the waveform after activating it to zoom out enough to see if the tops are flat.

http://www.digitalfaq.com/forum/atta...udio-levelsjpg

I don't think the graph shows every sample, just every nth sample, so I think the graph can only confirm that there *is* clipping, not that there *isn't* any.

In Virtualdub, select Volume Meter in the Audio menu. http://www.digitalfaq.com/forum/vide...html#post45240

You can view this in Preview mode before capture. Or, I think it can be viewed during capture in some cases. I have an ATI AIW capture card, and it already happens to be that I must be in Preview mode for capture to work. So I get it for free when capturing. Not sure if it works In Overlay mode. But, you may be able to capture in Preview mode without any negative effects. Experiment and find out. If not, you can preview the loudest parts of your tape before capture.

In Avisynth, command Histogram("audiolevels") shows audio levels as video plays. Refer to http://avisynth.nl/index.php/Histogram.

Sound Forge (not free) I know shows dB level along with the whole waveform. Audacity (free), I think would do that also. From virtualdub, load the video file, then save as wav and load that wav to one of those programs.

The meters on the AG-1980 (or any other device) can help but only after you understand the levels and limitations of each device through out the signal path, including the effects of any audio compression, limiting and automatic level control in the signal path. Clipping can be hard with flat tops and soft with somewhat rounded corners on the flat tops.

To make matters worse there is little consistency in published specifications among brands in the world of consumer electronics. And a pass-thru signal may be treated differently from a record/playback signal.

Clipping can happen at any stage in the audio path, and may not be evident on meters. The most reliable way to assess on the fly when shooting is with good headphones. In post opening the audio file in audio works station software like Audition will allow you to examine the waveform in detail and spot clipping.

Audacity has a feature to display clipping. https://manual.audacityteam.org/man/...#Show_Clipping

But this only shows samples that hit 0dB. In ELinder's screenshots, it looks like there is clipping despite having headroom.

If you install ffmpeg and point Audacity to it, you can drop any video file into Audacity and it will read the audio track(s). https://manual.audacityteam.org/man/...r_windows.html

Thank you very much for that information. In Audacity, I was zooming in all the way and didn't see any flat tops, I only saw them when I was zoomed out and getting confused. Now I know that the clipping was in the analog domain, not digital.

I've connected in a volume control to dial back the volume slightly, and it looks better on the waveform in Audacity, and it also sounds better. But I wonder if a pair of these attenuators would be better? Less noise, easier to reproduce the same audio capture levels every time instead of trying to find the same place on the dial?

Also, is it a good idea to normalize the audio after capturing as standard procedure?

In general, I do this, though only when restoring all of the audio at once.

And it really depends on the current overall loudness, as well as existing spikes. Normalizing isn't necessarily tied to loudness (nor volume, nor gain), merely adjusting levels to %. And that % is often arbitrary, being that it's not actually tied to loudness/volume/gain factors. Though it does affect volume/loudness (still not gain, as gain happened pre-capture).

Confused yet? :laugh:

I usually do 60% normalize -- but ONLY if peaks adjust upwards, not shrink. If shrinkage happens, you have a clipping/distortion issue to address.

BTW, the original issue in this thread sounds (pun not intended!) like a gain issue, either at recording time, in VCR, in capture card, or some other optional workflow device. That happens. Tracing/troubleshooting is required.

FWIW: I have encountered clipping that happened in the first stage of a recorder audio input. The input was set for a low mic level as with dynamic mics (typically -60 dBV output) but the connected mic was a self-powered condense mic (typically -35 dBV output). Thus the input signal was 25 dB higher than expected for the input configuration. Level controls that were located after the input stage were adjusted for normal VU meter indication with only occasional peaks above -10 dB indicated.

Later examination of the waveform (using Audition) showed clipped waveform peaks much like those in the graphic earlier in the thread. The clipped peaks were perhaps 6 dB below max digital record level. The problem was over driving the first input stage due to an incorrect setting of the recorder audio input.

It boils down to knowing what is happening in the audio chain during the initial acquisition both when making the original recording, and when capturing if from an analog source.

I just did a quick check on the audio output levels of the AG-1980.

With a 1 kHz sine wave recorded on a stereo track an output meter indication of 0 dB corresponds to about 0.3 volts open circuit. The meter's highest indication is +10 dB and that would correspond to about 1 volt RMS.

In light of the earlier post about the BM Intensity audio levels, audio where the AG-1980 playback meters go into the red (above 0dB indicated) will likely be clipped if using the Hi-Fi setting on the BM Intensity capture device.