I have Audacity 2.0.6 (.exe) and WIndows 7. I also have Behringer UFO202 A/D USB. This works normally as expected when connecting my PC to my amp (Marantz). For recording to Audacity to transfer some of my old cassettes from my NAD 613 deck, I have to set the record volume in Windows / Audacity very low. It’s at 0.04 in Audacity which is 4 in Recording Device / Advanced / Level.
It all works OK but I am concerned the level has to be set so low. Doesn’t seem right. The NAD deck Line Out is directly connected to the Behringer Line In and the Behringer USB connected to my PC USB as normal. (And yes, I have checked that the Behringer is set to Line and not Phono).
That is normal.
For some reason, Microsoft thought it a good idea to set +30 dB of software gain by default (amplifying USB audio input by +30 dB). Lowering the recording level to “4” (in Recording Device / Advanced / Level) is about right to set the gain to unity (straight through - no software amplification).
I don’t think that should be regarded as confirmed or official without more evidence from Microsoft itself that they (rather than OEM computer manufacturers or USB port manufacturers) are responsible. I have not researched it. Perhaps Microsoft give “guidance” to OEM’s about this in the same way they give “guidance” that it is a simpler user experience not to include stereo mix?
It is notable that some users find differences in levels between different USB ports on the same machine. I don’t know if that is to do with USB 2.0 and 3.0.
As I said in reply to the developer concerned, not everyone on Windows Vista and later experiences what you suggest, even though many with USB line level devices may do so. I have three USB audio devices (USB sound card with line and mic inputs, Ion USB turntable and Griffin iMic with line/mic switch) that all give “about” the same level on Windows XP and Windows 7. The level is a little higher on Windows 7 but my recording level there is usually about 0.5 on the Audacity recording slider.
There is “something” going on because USB microphone owners sometimes say they get a good level on Win 7 but have to get closer to the mic or use boost on XP. But others say there is no difference.
Clothears, if you set Audacity to 0.01, do you get a lower recorded level than if you set Audacity to 0.04, and higher recorded level if you set 0.07? Make that setting in Audacity, not Windows.
And is your USB port light blue, indicating USB 3.0?
Yes, if I change the level in Audacity, the record level varies accordingly. And the level reported by Audacity, gets followed by the level reported by Windows - that is, if i set 0.04 in Audacity, the Microphone recording level in Windows is 4. I just checked - leaving Microphone Levels window open and moving the slider in Audactity changes the number in the Levels window in Windows (if you see what i mean).
all my ports are USB 3.0 (with blue thing in the port).
While I am recording, I don’t use the Marantz, I connect the NAD directly (Line Out RCA) to the Behringer Line In RCA and the Behringer’s USB lead is in a USB 3.0 port. (I bought the Behringer to start experimenting with external D/A and also to record - now my usual D/A connection from PC to Marantz is Merdian Explorer.)
But if setting the record level so low (0.04) is a normal thing as advised by steve, then I stop worrying about it!
All USB “line level” devices that I have used (quite a lot) produce “full scale” (+/- 1.0) digital output for around 0 dB analog input.
Many (most?) USB audio devices use generic Windows USB audio drivers, and we have very many reported cases where the recording level in the Windows control panel need to be set to a low level to achieve a 0 dB digital signal from a 0 dB analog input. There is only one conclusion, which is that for default settings in the Windows control panel, the digital input is being scaled to a much higher level. When that is with generic Windows drivers and the Windows sound system, who else is responsible other than Microsoft? The A/D converter simply produces a digital signal in the range +/- 1.0. Whatever happens to that digital data once it enters the computer has nothing to do with the hardware device.
That’s what I meant by “normal” - it happens with many USB line level devices.
My assumption is that Microsoft do this based on the idea that most users will be using cheap USB microphones that lack adequate pre-amplification of the microphone signal prior to A/D conversion. In such cases a high level of digital gain will be required, so I don’t think that Microsoft’s decision is wholly unreasonable.
Of my USB audio devices, only the sound card provides manufacturer’s drivers - the other two devices use standard Windows USB Audio Class drivers.
I have just now uninstalled the sound card’s manufacturer’s drivers (so it’s using Windows drivers).
The Windows sliders for the line and mic inputs were reset to 100. Is that typical? I never really paid any attention to it, because I would not record at 100.
I adjusted the Audacity recording level to 0.5 where I had it before and re-recorded a known passage from the ION turntable into the line-in of the soundcard. It looks like it’s about 2 dB louder than using manufacturer’s drivers. It still doesn’t reach full scale.
I have now put the manufacturer’s drivers back, and the levels were reset to 80.
There are quite a few reports that the same USB ports on a given machine produce much quieter input on XP than on later versions of Windows. I don’t see that, so clearly hardware or drivers has some connection with it too.
I think all you can say is that this problem only got frequently reported after Vista came out. But for most users their Windows version is tied to their machine, so even there it could be a hardware/OEM issue to some extent.
Exactly - it is an assumption, and yes it has some logic to it. But you presented it as a hard fact with an exact dB modification level based on one developer’s report with two cheap (un-named) devices.
Are his devices applying their own internal unconfigurable positive gain? Are mine all applying their own unconfigurable negative gain? How else do I explain that on Win 7 or Win 8 I can record a line level USB signal without distortion at 0.5 recording level with three different devices?
Yes that should happen. The developer we’re taking about said Audacity can’t set recording levels between 0.1 and 0.
I don’t see that. With all my USB devices I can set 0.09 in Audacity and the Windows slider moves to 9 and I can set 0.01 in Audacity and the Windows slider moves to 1. 0.01 appears to record at a lower level than 0.09 if I record a sine tone, but the recorded signal at either level is far too low.
Mine are all USB 2.0.
I think we can at least agree that it is a common problem I have not had that issue with my USB devices on Win 7, either using my current laptop or the desktop I had before that.
My “assumption” is regarding “why” Microsoft decided to do it.
The “hard fact” is that it is common (not “abnormal”, aka “normal”) that USB devices are subject to a high level of software gain on Windows. From numerous reports on the forum, around +30 dB appears to be a ballpark figure for the amount of gain that is commonly applied. I was not saying that it always happens, or that it is exactly 30 dB, but rather I was assuring clothears that it is not at all uncommon on Windows machines and it most likely not due to anything that he has done wrong.
Most USB audio devices receive and send audio data only, and have fixed gain A/D converters. Typically, if the gain is adjustable, it is done by changing the gain of an analog pre-amp prior to theA/D converter by way of a “gain” control on the USB device. A typical (though now dated) integrated A/D with USB interface used in USB sound cards is the Burr Brown PCM2900. As is shown in the spec sheet, the maximum input voltage is 0.6 V p-p, which is a little below standard “line level” for consumer equipment, thus requiring a small amount of negative gain prior to the A/D converter to achieve 0 dB digital output for 0 dB analog input. As with most integrated A/D USB chips, there is no mechanism for controlling the gain via software. Thus, the gain of “the device” is independent of what computer it is plugged into.
It’s the onle A/D I have (other than the mic in on the Dell) so I can’t do any back-to-backs as Gale can. But the behaviour I see is as described by steve. I know the Behringer is a) not exactly up-to-date and b) was inexpensive (£25) so did not expect anything other than a useful experimental / get used to digital audio gadget. It has done exactly that for me and I’ve been pleased with the quality (it proved to me that a reasonable external device can be better than the onboard).
I was never suggesting clothears did any thing wrong, nor am I denying the problem is common.
Only until it’s plugged in. We already have evidence that plugging the same hardware into different USB ports on the same machine makes a difference, and that plugging the same hardware into a computer running XP will likely provide less gain than plugging into the same computer running Win Vista or later. I see that, but only to a small extent.
The XP versus Vista and later distinction could be due to Microsoft (assuming it is not something to do with more recent USB ports or more recent computers).
Then there is the question of drivers that may or may not be installed. Those will or may apply gain to the digital signal, is that not correct?
Looking at my own USB devices, only one of which has a gain knob, two of which can only use Windows USB class audio drivers, it makes no sense that Microsoft are applying +30 dB gain that I can’t opt out of. That would imply that all three devices are set to very considerable negative gain prior to the A/D - and as all three devices are of XP vintage I very much doubt that.
As I said, I also don’t see that Audacity is unable to set levels between 0 and 0.01 that have an effect on the resultant audio. Do you see that?
Or just after.
Typically we’re taking about 16 bit serial data at the USB socket, and that bit is the same regardless of the computer. Then the data is subject to drivers/software control, and at that point anything can happen, and we see a lot of variability with required user control settings. What we don’t know yet is how much of this variability is in kernel mode and how much in user mode. Perhaps “WASAPI IAudioCaptureClient” will provide a more consistent user experience, if Audacity gets that ability in the future.