I’ve been capturing audio with a USB ADC device, then through ffmpg, then transfering to a windows machine. When I open the file in windows, it looks like waves are hitting the ceiling.
This image shows two similar tracks --however one of them should have been less ‘loud’ during the capture. I’ve been focusing on adjusting gain from the capture digital device and when that seemed to make no difference, from the preamp analog volume device (before it feeds into the ADC). (The preamp does work when outputting to an amplifier. But it seems to not make any difference when capturing to file. Looking for a sanity check and now looking at if this is an audacity feature. I use audacity just to clean up the start and end locations and add basic metadata.
That’s a bit confusing… Are you recording (capturing/digitizing) with Audacity? I assume this shows-up before FFmpeg gets involved…?
What “device” are you using, and what’s the source? (a microphone, etc?)
Assuming that’s an analog knob, that should work. Does the volume adjust down to zero for a silent recording?
With a USB device an ANALOG adjustment HAS to be used. The analog-to-digital converter in the soundcard/interface will clip if you “try” to go over 0dBFS, and lowering the volume later digitally doesn’t remove the distortion.
…If you can eventually solve the problem and adjust the recording level, low digital levels are not a problem. Leave some headroom. Amplifying digitally later is a lossless process. Pros record at very low levels (-18 to -24dB). That’s not really necessary unless your levels are unpredictable. Typically, for digitizing analog records or tapes (which are pretty predictable) it’s recommended to shoot for peaks between -3 and -6dB . But nothing bad happens when you get “close” to 0dB and lower levels usually OK.
Thanks for the feedback! Now I see the passive preamp was connected further down the chain, so adjusting it had no impact. So that was it. on the ffmpeg filter, there is an automated -filter:a “volumedetect” report, however I do not think it detect levels above 0db (this is on the same input average) (it should list max volume above 0 but it must clip at 0 then…):
[Parsed_volumedetect_0 @ 0x55b5a663c0] n_samples: 17384172
[Parsed_volumedetect_0 @ 0x55b5a663c0] mean_volume: -11.6 dB
[Parsed_volumedetect_0 @ 0x55b5a663c0] max_volume: 0.0 dB
[Parsed_volumedetect_0 @ 0x55b5a663c0] histogram_0db: 100499
One last question. I think I found the sweet spot at “volume=-1.3dB” . But when I’m looking at audacity, I can’t tell if this is a clip from the capture or a clip from the final mix. The source is a vinyl record. Is there a way to line up the graph ruler with the apparent clip in this graphic.
If that’s just a reporting filter then it’s just confirming that you are clipping at 0dB.
FYI - Audacity can Show Clipping. I(I don’t remember if that’s enabled by default).
But it’s only checking the peak level. It’s not checking the wave shape and you can get false negatives or false positives.
If the waveform is clipped and you lower the volume with a negative Amplify setting the peaks cab be brought down below 0dB and it will no longer “show red”.
Or if the wave isn’t clipped and you boost the bass (or anything that boosts the level) you can push the peaks over 0dB it will “show red”. But Audacity uses floating point and it can go over 0dB so the wave isn’t really clipped (yet). You can lower the volume before exporting and everything will be OK.
Are you saying that if I record with audacity rather than ffmpeg, it will preserve the wave without clipping and allow me to adjust levels before saving? That would be nice, since I have to deal with the adjustment before recording otherwise… unfortunately the computers attached to the stereo are headless linux based and do not have a GUI.
I guess I’m not sure what’s going on yet. After setting the capture with a -1dB filter, I still see flat lines in audacity (as a sort of visual assessment). So all I appear to be doing when applying a filter is shrinking on average the volume from 0 dB to -1db — it appears the flat line is in the source or that I’m still creating the flat peaks but at a lesser volume? This is happening on many records and wondering if the phono preamp is causing this.
Alright, so for anyone actually reading this, sorry to spam.
I looked in the forum and decided the best way was to select the entire ‘side a’ and to normalize after removing a couple of ticks. Pretty happy for a quick job but if there’s a better way, woudl love to know it.
I Figure I’ll end this thread with the three versions that literally just helped me realize what I’ve been dealing with without knowing until now about my phono preamp.
The first clip is with my phono preamp’s lowest 30dB gain setting. The second clip is the first clip with the audacity normalize plugin, and the third is with the phono preamp’s 42 dB gain which I always had as standard. Crazy stuff.
