On Sat, 8/2/2025 10:06 PM, J. P. Gilliver wrote:

I'm currently transcribing some old records (33s and 45s), and having
trouble persuading the PC to let me hear the input.

I'm using a USB-powered phono preamp; that works fine. It appears in
Sound | Recording as 2-USB PREAMP, and the OS thinks it's a microphone.
Yes, I _do_ know about the default being off for microphones to prevent howlround, and under Properties | Listen for the device, I _have_
changed "Listen to this device" to ticked.

Trouble is, in GoldWave which I'm using to do the capturing, I can have "Monitor input on visuals" set (otherwise the level meters don't move!),
but as soon as I do that, it no longer comes through the speakers. (I
can still record, but it's a lot easier to do so if I can hear the
record playing!) It seems to be a "sticky" thing in the OS - at one time disabling and then re-enabling the device (somewhere in Windows
settings, not GoldWave) made it cone out of the speakers again, which suggested to me it wasn't a GoldWave problem.

Anyone any ideas (other than "stop using Goldwave")? Anyone had similar problems in keeping external (especially USB) sound input devices
audible? If so, what did you do to get round it?

Windows 10-64.

Right now, your hardware is relying on GoldWave to handle all the recording and output chores.

Phono-2mV ------ USB stereo 44.1KHz 16bit ADC <=== Goldwave recording here

*******

Logically speaking, the (pseudo) analog path through a PC, has an "input multiplexer"
which samples microphone and line input. The output can go directly to the output
speakers (passthru, possibly at least one application like Goldwave as App1 would
need to passthru, to reach the output mixer). Which might be viewed in some way,
as the "Monitor" output. Then, using the "Stereo Mix" option (a hardware path in the HDAudio chip), you can record the Stereo Mix ("What you hear") and digitize that.

| \ (SSO is system sounds output, "ding")
--- | \ input mux as "mixer" App1 -- | \
--- | \___--------------------- App2 -- | ---- LineOut
--- | / SSO -- | /
| / | /

You might take the analog output of your USB Preamp, and feet that into
the HDAudio on your PC, with the intention of recording off the StereoMix recording path on the HDAudio.

Phono-2mV ---+-- USB stereo 44.1KHz 16bit ADC --x
|
+--- HDAudio-Line-Input ----HDAUDIO-chip -------------+-- LineOut --- [HeadPhones as clean monitor]
|
+-- HDAUDIO-ADC --- GoldWave | (StereoMix AKA "What-you-hear")
| | is what you select for recording.
+-----------------------------+

I find this stuff very confusing, when as a user I'm expected to "discover"
how the PC is supposed to work, and make my applications work properly.
This isn't a Jack subsystem, ASIO, M-Audio cards with more inputs
and outputs and a mix-down architecture we're doing here, we're <cough> "abusing" the limited Windows system audio architecture to make
our little recording studio. And it is GoldWave that *must* have
some model in mind, when it promises to allow you to Monitor and Record
at the same time. We can't "force" GoldWave to work properly, or any other application (Audacity) for that matter.

When I work with Audacity, I can't expect to do Input and Output simultaneously with the one application instance. Sometimes I get to hear what I'm doing,
and via the echo suppressor in the driver, it doesn't howl (too much).
I've had to use two applications, and fiddle about with cables and the Stereo Mix,
to find "some way" to do a thing. It's not exactly "convenient", whatever you end
up with in the end. It would not be out of the question for example, to end up using
*two* PCs to achieve an objective. Or *two* applications, one for Record, one App for
Playback or monitoring or something.

As a bit of a joke, I tested Miracast a while back. Miracast normally allows displaying a PC video output, on a TV set (uses Wifi), and it requires some coupling
between the GPU and the Wifi to do that. Well, they made an option in
windows, so a PC can *receive* a Miracast screen and display that stream
on one of the display panels. And I set up a second PC as my Miracast receiver (because I don't own a big-ass TV set!). And it worked fine -- this likely would not have worked, not that many years ago. But it worked, and it was
kinda looking like a Remote Desktop Protocol instance, but the mouse
didn't work :-) That's an example of abusing some Windows functions, although the function achieved in that case, is not all that impressive. But I was happy, because I got to "pipe clean" the Miracast concept, and see a less-integrated set of hardware drivers actually work together to
achieve a result. Previously, the solution was semi-proprietary and
was a "reward for buying Intel or Nvidia or something". It originally
wasn't all that generic, and only certain specific hardwares could
be expected to work together.

The audio isn't quite that bad, but the history of StereoMix is quite cloudy. At one time, a number of goofballs were claiming "StereoMix no longer works!". But from my point of view, if you looked at all the HDAudio chip internal
block diagrams, the path was there... and all that we needed to do,
was use an OS that still honoured the hardware details and tap into that.
In some cases, there was a "Recipe" to make this input "re-appear" :-)
Hideous concepts. There is no guarantee, that your setup will look
like this, but with enough voodoo, you can bring it back. See, StereoMix
still exists. Here is the picture. The Win10 machine across the way will
also have one of these.

[Picture]

https://i.postimg.cc/zGy0FYz5/Loopback-Stereo-Mix-As-Input-Choice.gif

Some of these things would be more apparent, if we could look at the
RealTek proprietary control panel for our RealTek audio chip, but
Microsoft has replaced the RealTek on windows 10, with their "Generic" simplified panel. But I still have at least one SSD with a copy of Win10 on
it, where the RealTek panel is still there. I don't know why that
hasn't been zapped and killed. I think currently, the RealTek driver
has been containerized, similar to how the NVidia GPU driver has
been containerized, to "protect" the OS in Ring0 from some sort of attack.

Summary: It's a miracle, if anything still works on these machines :-)
I know there are devs, working somewhere, with good intention,
trying to make stuff work, but it's an uphill battle, like
swimming in sewage as a hobby. I have some instances of stuff
that *still work*. "Take that and that, you bastards" :-)

Paul

--- MBSE BBS v1.1.2 (Linux-x86_64)
* Origin: A noiseless patient Spider (3:633/280.2@fidonet)