Win7: MME /vs/ Windows DirectSound?

I took a break for a while, upgraded from Vista to a clean Win7 installation, still using Audacity 2.0.2 and a Behringer USB input from the tape deck. And hoping this change of OS gets rid of the mysterious noises I’d been getting no matter what I tried under Vista.

Now I see Audacity’s preferences gives me a choice of two recording “hosts”, WIndows DirectSound which appears to be the native USB audio, or “MME”. And the volume on the DirectSound is incredible, with the Win7 input level set around 5 I still need to back down the level in Audacity below 0.1!

Should I assume this is normal with Win7 and stick to the DirectSound USB choice?

Anything else a latecomer to Win7 should be aware of for Audacity?

–Red

See this post: https://forum.audacityteam.org/t/mme-or-windows-directsound/17434/3

I have Win 7 64 bit on a pretty fast machine using the latest Audacity. I find that MME gives me better performance (and somewhat better sound) over Direct Sound. Of course I had to tweak the latency when I first installed it, but it’s fine and I see no reason to change. Am I missing something?

Thanks, Steve. that post of Gale’s makes me think DirectSound is the way to go.

Problem with this is the volume. I’ve got the control panel settings configured for a “mic” volume of 3, on a scale of 1-100, and in Audacity I’m still seeing clipping with the mic level set below 0.1. Reading the FAQ, I see I can double-click (I won’t call that intuitive) to get a pop-up window and dial that in, but there’s no other way to get better granularity out of the input volume, than to pop a window open and closed?

I’m just leary of setting the control panel volume level so close to zero. There’s no problem, no loss of anything, if I push that down to something insane like a “1” ?

And, if I have the Audacity monitor set to “linear” so it is reading in tenths up to 1.0, does that correlate to the mixer settings? i.e. if I’m showing maximum input levels on the monitor peaked at .7, and my input is set to .5, does that mean I could raise the input level two or three tenths, so I’m recording with a peak just below 1.0?

I use Windows Direct Sound on my W7 laptop - when I use it to record streaming audio I too find that I have to crank down both the mic and the spejer levels a long way in order to get a sensible input signal level peaking at 50%/-6dB. This seems to have no impact on the quality of the recorded sound.

I chose WDS over MME after some discussions with Gale - I chose it not for its allegedly superior latency (I don’t need to take account of that for LP and off-air capture) but as it was the more modern implementation from Microsoft.

WC

WDS over MME, right with you on that. I kept reading in the FAQs and latency is a total non-issue for “straight” recording from a single source, as opposed to mixing.

I think I’m down to “1” on the 1-100 scale in Windows, and something like 0.65 on the input slider in Audacity.

Incredible to think a “simple” software change has a 100x effect on the volume.
I suppose that’s like complaining that whenever my dog sings the star spangled banner, he’s off key.

Good boy. Cookie?!

WDS is not the Behringer USB, and MME the onboard sound. WDS and MME should be able to access both devices.

I have not heard before that WDS had higher levels than MME (I have not experienced it on different Vista or Windows 7 or Windows 8 systems). When you talk about adjusting the mic slider is this for the Behringer or the built-in sound, or both?

If it affects built-in sound, have you already looked into whether you have the best drivers for the sound device ( Missing features - Audacity Support) ?

Really, the Audacity input slider and the Windows slider “should” be moving in tandem for the same input device (that is, move one slider and the other moves, so they are both at the same volume). If that is not the case, the Audacity input slider is supposed to “grey out” so you have to use the Windows slider instead.

You can get reasonable granularity using the mouse to drag the volume sliders - the problem is just that the tooltip rounds the volume to one decimal place. To get more control, hold SHIFT while dragging. This applies to the Track Control Panel sliders too. You’ll see you can drag the slider much more slowly (but there is a quirk that the mouse pointer pulls away from the slider).

The 1 to 100 Windows granularity could be specious (and the Audacity five decimal places granularity in the dialogue box almost certainly is specious, assuming the Audacity and Windows sliders are moving in tandem). It’s up to the sound device what granularity it actually supports.

I don’t think you can draw any direct conclusions at all if the Audacity slider and Windows slider are at variance.

If they were not at variance, a slider setting of 0.5 doesn’t mean you will record at 0.5 on Meter Toolbar, or 1.0 means you record at 1.0 on the meter.

It “should” mean (I think) that reducing the input slider from 1.0 to 0.5 will halve the input level from whatever level it was at. So if 1.0 on the slider was achieving 0.5 (or -6 dB) on the meter, 0.5 on the slider should achieve 0.25 (or -12 dB) on the meter.

@Peter - WDS in my experience doesn’t usually have (much) less latency than MME on Windows Vista or later, because on those versions of Windows, MME and WDS are both emulated. On XP, WDS has more direct access to the hardware than MME does (as in “Direct” in its name), so should be faster than MME.


Gale

if you choose Windows Direct Sound (WDS) and you have both “exclusive mode” boxes checked in Windows, Audacity should be able to request the sample rate (as set in Project Rate bottom left) direct from the device, without Windows intervening to do resampling (which it would under MME if you choose a rate the sound device does not support).

This does mean though, that if you do request a rate not supported by the sound device, you could get bad sound or speed problems.

This is usually more of a problem with external interface devices (which may only support a few rates) than built-in sound devices.

This comment about Exclusive Mode only applies to Windows Vista and later. It’s complex but you can read more about it here:
Missing features - Audacity Support .

MME is generally more compatible than WDS, even on Vista or later, which is why it’s default.


Gale

Hi Gale.

WIndows DirectSound which appears to be the native USB audio

WDS is not the Behringer USB, …
When I said “native USB audio” I meant native to Windows7. Not device-specific to the Berhringer, just that the path/drivers/subsystem seem to be the new default for USB audio in Win7.

I have not heard before that WDS had higher levels than MME (I have not experienced it on different Vista or Windows 7 or Windows 8 systems). When you talk about adjusting the mic slider is this for the Behringer or the built-in sound, or both?
This is for adjusting the microphone slider in the “Levels” tab for the mic in the Control Panel. Sound, Recording, Two “microphone” items are listed. One for the Soundmax integrated digital HD audio, the other saying “USB Audio CODEC” so again, I don’t think either one is referring to the Behringer as hardware so much as they are referring to the audio path in Win7. One using the native USB codec, the other using presumably the USB channeled through the older audio routines. Either device WILL work depending on what settings are made in Audacity, there’s just a huge difference in the volume they produce.

If it affects built-in sound, have you already looked into whether you have the best drivers for the sound device ( Missing features - Audacity Support … ce_Drivers) ?
This is a new installation, clean installation and total update of everything, about a month old now. I haven’t looked at the wiki docs since “everything” is theoretically up to date.

Really, the Audacity input slider and the Windows slider “should” be moving in tandem for the same input device
Well, neither of the Windows devices would allow for a level setting below “one”.

You can get reasonable granularity using the mouse to drag the volume sliders - the problem is just that the tooltip rounds the volume to one decimal place. To get more control, hold SHIFT while dragging.
Thanks. I did get that from the FAQ after browsing around the different entries. hadn’t seen it before because I was looking for “volume” (no entry) rather than “level” etc. Actually I’m doubleclicking the slider so I can put in a numerical input and that’s working nicely, although I wish there was some way I could see just how much clipping was occurring, rather than guess at how much to lower the input to stop it.

It’s up to the sound device what granularity it actually supports.
Well…nothing references the device itself as hardware, really. I think in this case it is a question of which audio subsystem is being used (i.e. the USB codec with or without the old emulation layer on top of it?) because nothing is actually adjusting the Behringer, the volume adjustment is happening in the drivers, which are part of Win7 AFAIK. I could easily be wrong about that I’m just going by appearances.

I don’t think the OS presents the devices any differently, because it has no idea what API the recording device is going to call. USB Audio CODEC must be the Behringer if that is the only USB device connected.

Are you asking for the Audacity meters to represent audio above 0 dB?

Both WDS and MME are emulated on Windows 7; those are the only choices Audacity has until it supports WASAPI.


Gale

Gale-
"Are you asking for the Audacity meters to represent audio above 0 dB? "
If that’s physically possible that would be one way to make it happen. Or if not can I:
1-make a recording, note the highest volume is maybe 0.8,
2-and then increase my input level from whatever it is by an additional 0.19,
3-and then make a new recording and expect it will show the highest volume at 0.999 instead?

In other words, measure max volume, do some real simple math, redo the recording and see the volume maxxed out without clipping?

“Both WDS and MME are emulated on Windows 7; those are the only choices Audacity has until it supports WASAPI.”
I like wasapi on my sushi. [sic]
I’ll look forward to Audacity 3.0 and meanwhile keep looking for ways to get the noise out of 2.0.2. (Yes, the “static” attacks I had under Vista are still here in Win7. Different sound, but same problem. Perhaps if I installed the OS and the app in a month that didn’t have an “R” in it…Oh, wait, no, that’s just for bad oysters, isn’t it? )

I don’t think it would (even theoretically) work like that because 0.8 on the meter could relate to different settings on the slider. It might work theoretically like that if your 0.8 on the meter was caused by 0.8 on the slider. If 0.8 on the meter is coming from 0.1 on the slider I would expect you to need much less than another 0.19 on the slider to get to 0.99 on the meter.

I see a clear relationship as I described between slider and meters when recording a tone in Audacity using stereo mix and adjusting the output slider in Audacity; for example if 0.5 on the output slider causes 0.8 recording level on the input meter, 0.25 on the slider causes 0.4 on the input meter.

Recording the same tone, the relationship with the input meter is not quite linear when moving the input slider; if 1.0 on the slider causes 0.5 on the meter, 0.5 on the slider causes 0.3 on the meter. And if I record the tone from the loudspeakers using a mic, I have to move the slider a lot further to get an equivalent change in input level.

I still don’t understand for which physical (hardware) device you are trying to control the input volume. Is it the Behringer with the tape deck attached to it, or a mic attached directly to the computer? Can you make test recordings to get the level correct? That is the best practice.

This is recording from the Behringer, and Windows Sound Recorder has the same problem as Audacity?



Gale

Thanks for bringing this up hellosailor - it gave me a useful nudge to update the Manual with the additional insight that Gale gave me earlier in this thread. :slight_smile:

See:
http://manual.audacityteam.org/man/Device_Toolbar
http://manual.audacityteam.org/man/Devices_Preferences
http://manual.audacityteam.org/man/Tutorial_-_Selecting_Your_Input

Peter

Gale-
"I see a clear relationship as I described between slider and meters when recording a tone in Audacity using stereo mix and adjusting the output slider in Audacity; for example if 0.5 on the output slider causes 0.8 recording level on the input meter, 0.25 on the slider causes 0.4 on the input meter. "
So you’re saying it is a proportional relationship, that is, a 50% change on the slider brings a 50% change in the level recorded. Not a linear 1:1 correlation on the scales, but the change in proportion carries across as a “same” change in proportion. Logical, I guess. Something to consider changing, or building an alternate choice, so that a change of “0.3” up or down on the input, caused the same change of “0.3” on the output. Whether it was on the linear scale or the db scale, either way I’d have expected “same same” the same way that a 3db increase on a mixer would cause a 3db increase on recorded volume.

“And if I record the tone from the loudspeakers using a mic, I have to move the slider a lot further to get an equivalent change in input level.” I’m afraid to guess, but maybe that is because the mic is not necessarily a linear device? Although of course playing and recursing a test tone would allow the software to MAKE the settings linearly proportional, that wouldn’t be a trivial project.

“I still don’t understand for which physical (hardware) device you are trying to control the input volume. Is it the Behringer with the tape deck attached to it…?”
Yes. The Behringer has no volume adjustment, and the tape deck I’m using has not output level adjustment. The resulting output from the deck, or from the Behringer, “is what it is” and there’s no adjustment possible. I know, that’s not professional but my Nakamichi needs some expensive repairs so I’m using what I can use, and relying on the PC’s software to make recording level adjustments.

"Can you make test recordings to get the level correct? That is the best practice. " On the Nakamichi, sure. But to make test recordings from the tape deck into the PC/Audacity? There’s no such thing as a test, all I can do is make a recording, watch it, and if it goes into clipping, dump the recording and start over again with a guess about how much to lower the gain. So that can cost 1/2 hour to an hour every time there’s a surprise, and if I haven’t guessed right or want to make it closer, that’s another 1/2 hour to hour lost. It can make a one-hour transfer into a six-hour task.

MOST of my tapes were carefully made on the Nakamichi but since they were made on three or four different tape types (two metal, chrome, high-bias) each of them also has some different characteristics, even if the music was similar. So I get batches that run fine with one pass, and others that make a simple job into a much longer one.

Being able to analyze the music and just match up some numbers would at least mean I didn’t have to be there watching and guessing.

Peter-
I’ve had the toolbars turned on but part of the problem is simply that I’m never quite sure (bear in mind, I just upgraded the damned OS last month and got Aud back online a week ago) fo some of the selections to make. Sometimes, it seems that I can’t get the computer to play what is being recorded no matter what I set, restart, or change. And yes, I have play-through enabled. Then sometimes, there’s no telling why two different choices work, sometimes the same, sometimes not the same, no real indication which is better. And why the buzzing came back the other day, I have NO idea but there’s probably something that was wrong with one of the selctions I was trying.

In Audacity:

First choice, WindowsDirect /vs/ MME, apparently should make no difference in anything except latency (not an issue here) and volume level (again not an issue, either can be adjusted.)

Second choice, Speakers, Primary Sound Driver / Soundmax Integrated / SPDIF Interface (SoundMax) / Speakers (USB Audio Device)
and one would think the Primary Sound Driver, or Soundmax Interated, would both be the same physical device, the laptop speakers. I’m not sure what the last reference is to since there are no USB speakers plugged in, this is pointing back to the Behringer’s capability to have earphones I think.

Third choice, recording device, Primary Sound Capture Driver / Microphone (USB Audio Device) is again confusing since there is a physical mic in the laptop (and in the control panel choices) and no physical mic in the USB device, which is the Behringer.

Maybe Windows9 will include a recording applet that samples every one of the inputs and says “There’s activity on xx line, do you want to use that as an input?” and make all our lives easier. I don’t expect they will, I expect they’d rather make recording HARD to encourage purchases though.

Any advice for the overwhelmed and insufficiently caffienated?

I think this is probably out of Audacity’s control. For a start, both the Audacity input and output sliders are supposed to work in tandem with the system slider, for the device currently chosen in Audacity Device Toolbar.

The Audacity sliders are not supposed to be a software slider that modifies the Audacity input and output level without moving the system slider that controls the hardware. Of course the volume slider in Windows Media Player is just that, a software slider.

Indeed it is not professional… :slight_smile: Is there no working headphones output on the tape deck, with a volume control that attenuates that output?

It depends if you know the music beforehand, I suppose. I wouldn’t have that problem with the few tapes I have because I would just monitor the loudest part in Audacity then adjust the slider to reach -6 dB on the meters before recording.

Of course if the recorded level isn’t consistent on the same tape, that’s a problem.

It wouldn’t if you don’t know where the loudest part of the tape is and if the recorded level is inconsistent. Otherwise, you can make a test recording (or monitor before recording) and solve the problem.

In any case, whatever gain you need from 0.1 on the slider to reach a satisfactory level on the meter is probably repeatable with the same equipment.

Plus, you should not be recording up to 0.999 on the meter anyway, especially if you don’t know the music or the consistency of the tape level.

I may still have an old test build of Audacity that has an experimental “recording level control” made by a GSoC student. It is supposed to be used to automatically set the recording level in a pre-recording test, so you don’t have to fiddle with the input slider to find its optimum level. It is not supposed to be used for live recording, or it will probably act much like a compressor and limit the dynamic range. Do you want to try it, if I can find it? It will be an old Beta version of Audacity, unless the patch still builds, so entirely at your own risk.


Gale

All these symptoms are indicative of a built-in sound device that is broken (or its drivers are), and/or too close to the working parts of the computer and/or the computer is under too much strain.

What RAM and GHz is the laptop? What screen size?

The only playback difference should be this. If you have your speakers or headphones set as the default Windows playback device, then in Audacity you can choose either (a) that named speakers or headphones device, or (b) the mapped output device. The mapped device is either “Microsoft Sound Mapper - Output” if you choose MME host, or “Primary Sound Capture Driver” if you choose “Windows Direct Sound” host.

If you then connected the Behringer to the computer and made that the default Windows playback device (or it became such), you would lose sound in Audacity unless you had chosen the mapped output or unless you change the playback device to the Behringer.

If you don’t want the Behringer to take over as default Windows playback device when you connect it, make the laptop speakers default on the “Playback” tab of Windows “Sound”, then unplug Behringer from the laptop. Next time you plug-in the Behringer, the laptop speakers should remain Windows default playback device.

Note that this trick probably only works on Vista and later; on XP the Behringer would probably always take over as the computer playback device when plugged in.

Probably correct for Windows Vista and later. It would make a difference on Windows XP. Because of the lower latency (playback as well as recording) you could quite easily find on an underpowered XP computer that you only got software playthrough from a USB device when choosing Windows Direct Sound.

See above. Primary Sound Driver is only the laptop speakers when the laptop speakers are default playback device in Windows “Sound”.

If playback to the laptop speakers or headphones is unsatisfactory (in terms of predictability) why not use the headphones out of the Behringer, which is hardware playthrough? Then you can turn Software Playthrough in Audacity off.

The built in sound device can never be called “USB”.

Most USB devices that have no input selection choice in Windows “Sound” (such as Behringer UCA 202, USB turntables and USB cassette decks) will be called a “microphone” in the “Recording” tab of Windows “Sound”.

Note: If your Input Device in Device Toolbar really only has a choice of Primary Sound Capture Driver and Microphone (USB) then (at best) the built-in inputs are all disabled in Windows.

Please record from the Behringer in Windows Sound Recorder as an experiment (Windows Globe, type in the search box “sound recorder” (without quotes) then double-click “Sound Recorder” when it appears in the search results). You will need to make the Behringer default recording device on the “Recording” tab of “Sound” in Windows in order to make Sound Recorder record from the Behringer. Does that have static attacks?

If I was in your position I would verify the drivers of the built in sound device.

In Audacity for recording from the Behringer I would choose Windows Direct Sound host, set Audacity project rate to 48000 Hz (which is the usual Behringer specification), and check (tick) both “Exclusive Mode” boxes in Windows (see here for how to do that).

I would monitor the recordings in headphones in the Behringer and use the Behringer as the computer playback device (for playing songs in Windows Media Player and so on).


Gale