Audacity Forum

The only time I got dead zero on the meters was when I selected the S/PDIF input with nothing connected. Then it's a straight shot from the outside world to Audacity with no translations or other tricks. And no digital input. It didn't seem to make any difference where the volume controls were under those conditions.

All the rest of the times I was using one of the digitizers in or around the Mac. So it's quieter, but a perfect noise floor is not particularly easy for us, either.

I don't remember hearing from the linux contingent.

Koz

kozikowski wrote:I don't remember hearing from the linux contingent.

My last post has results for Linux.

OK Guys, I'm back. (Digression: had a good time yesterday evening; they even laid on a buffet supper for me!)

Following on from Steve's question about whether I was using Windows DirectSound or MME, I did some more tests. I first checked that MME was giving the same sort of results today as it did before - it was. I then switched to Windows DirectSound, reset the output device to Speakers (RealTek HD) and the Input device to Stereo Mix (RealTek HD) - i.e. the same settings I had with MME - and re-ran the test: i.e. recorded "silence" and then analyzed the frequency spectrum. It was a much lower level of "sound" (-82dB and lower compared to the MME of -60dB and lower). To double-check, I switched back to MME and tried again. Wow! Frequencies all the way up to 100Hz instead of up to the 10Hz that I had had before. I switched back to DirectSound - double wow! Frequencies all the way up to 1000Hz. I couldn't find a way of getting back to the 10Hz upper threshold again. In desperation, and uttering a silent prayer, I powered off and back on. Windows DirectSound was still the chosen option and I re-ran the test. I noticed that, at the very start of the recording, I saw in the levels meter the transient noise that I mentioned in an earlier post. I show below two frequency spectra: the first is from the entire recording (including this transient noise) the second is the same recording but omitting the first second of it.
I stand by my previous conclusions: the RealTek HD sound card installed in my desktop PC system, when used as the source device via Stereo Mix, is outputting some very low level, very low frequency sounds which Audacity captures faithfully. It does this whether via MME or via Windows DirectSound. As for that transient at the start of recording: it used to happen back in the days of analogue recording. The system would pickup the electrical and mechanical sounds associated with getting all the solenoids engaged and the spools turning. That's why we always waited a few seconds before starting to speak, remember? Nothing's changed in this digital age; we still need to let the kit settle down before we speak, sing, play or whatever.

Finally, that increase in frequency range each time I switched from MME to DirectSound intrigues me. It almost looks like some sort of programming bug whereby a decimal multiplier is not being cleared back to zero at each switch over. I have not done any further investigation into this. It may be that what I have given you will be enough for the developers to identify and correct the bug, if that is what it is. I am happy to dig further if needed.

regards,
Peter

Try this and see what you get!

Launch Audacity 1.3.14 using your normal settings. Press Record and record five seconds of "silence" then press Stop. Repeat to get another recording of "silence". Repeat again, and again. Do you see the level meters climb higher at each recording? I do! All those recordings should be pretty much identical but on my system they are not. Is this an artefact of my system or is it reproducible on others?

@Steve/@Gale,
The existing topic title seems to cover what I'm discussing but if you would rather I create a new topic, I'll gladly do so.

PGA wrote:Try this and see what you get!

Launch Audacity 1.3.14 using your normal settings. Press Record and record five seconds of "silence" then press Stop. Repeat to get another recording of "silence". Repeat again, and again. Do you see the level meters climb higher at each recording? I do! All those recordings should be pretty much identical but on my system they are not. Is this an artefact of my system or is it reproducible on others?

What do you have set as the recording input in the Device Toolbar?

steve wrote:What do you have set as the recording input in the Device Toolbar?

Stereo Mix (see attached screen-grab)

Do you also have "Overdub" enabled? (Transport menu)
If so, then each time you record a new track, other tracks will play through your sound card, and because you are recording from "Stereo Mix" they will be re-recorded into the new track, which will lead to a gradual increase in signal each time you record a new track.

steve wrote:Do you also have "Overdub" enabled? (Transport menu)

I did, but after reading posts on this forum I turned it off earlier this morning. I just didn't "put two and two together" and work out that Overdub was the cause of that little glitch. How does it feel being a genius?

"Overdub" is required if you want to be able to hear other tracks while recording a new one.
To avoid recording those other tracks into the new track, set the recording input to the actual input you are using (for example, "Mic" or "Line").
Of course, if you are actually wanting to record sounds that are playing on your computer then you will need to have the recording input set to "Stereo Mix", but then it will record other tracks that are playing.

The only time "Overdub" contributed to the results were the tests that I did and which I documented in the post that began "Try this and see what you get..." All the other tests were single recordings that were Analyzed, screen-grabbed and then the track deleted via the "x" alongside the "Name" in the Track Details panel. Please bear in mid that, in my normal workflow, I do not record using the computer. Everything is recorded on my Zoom H4 or is ripped from CD. I use Audacity only to handle digital sound files - not to digitize sound. My philosophy is that a computer was never designed to be a great sound recorder; but it was designed to be a great manipulator of digital data - and that is what I use it for! There are some, infrequent, occasions when I would like to capture sounds (from the Internet) that are already playing on my PC. Those sounds are already digitized - aren't they? - and so "recording" should just be a case of capturing that digital data - shouldn't it? I don't need Overdub turned on, and having it turned on was a combination of lack of understanding and oversight on my part. It is now turned off and will remain so.

I'm beginning to think that I might, unintentionally, have hi-jacked this topic. If that is the case, I apologise. The only reason I've been doing these tests is to try and help everyone understand what the source of some of the originally reported noise might be. I have, this afternoon, performed the same investigation into the recording of "Silence" on my Acer laptop, It is far noisier than my desktop system (-36dB as against -80dB and frequencies up to 40Hz as against up to 10-15Hz); so it is, perhaps, just as well that I use it only for playback of AV sequences and demonstrations of techniques for preparing the material for AV shows