[wdmaudiodev] Re: How to filter microphone voice?
- From: "Sam Tertzakian" <sam@xxxxxxxxxxx>
- To: <wdmaudiodev@xxxxxxxxxxxxx>
- Date: Thu, 25 Jun 2009 20:39:57 -0700
Hi,
If you want to change the audio for all possible streams/devices under 2000/XP,
then the best choice is a WDM "lower filter" driver that sits under KS.sys. The
driver will pass all requests unmodified. However, for reads and writes it will
modify the data. For writes, your driver will receive a WRITE IRP and will
modify the data (change voice) and then submit the same IRP with modified data
down to the next driver in the stack. For READ IRPS it will set a completion
routine, then pass the IRP down to the next driver in the stack. When the lower
driver has filled the IRP's buffer with audio, the IRP will arrive in your
completion routine. There you may modify the data that was read and then
complete the IRP which will send it to the application.
Keep in mind the following:
1. This architecture will work for 2000/XP/Vista/Windows 7. The same binary
should work.
2. This driver is a WDM Filter Driver. Do not confuse this with a DirectShow
filter which is unrelated to this.
3. Because it is WDM Filter Driver, it is not an "audio driver" per se, as it
is a straight filter driver. So, the information that is relevant in the DDK is
in two places: a) WDM Filter Drivers b) Kernel Streaming Drivers. You need b)
because the data format passed in the IRPs is passed from KS so you need to
understand the data formats in the IRPs.
I can tell you with certainty that all of what I am saying works, because I
have personally done this.
One more thing...what you want to do is not trivial. There are some details I
left out. For example, when you receive data buffers, they just contain audio
data. But, buffers themselves have no information about things like: how many
channels are being processed, how many bits/bytes per sample and the sample
rate. So, some of the requests that go through your driver, while they remain
unmodified as you pass them down, must be read by your driver and tracked so
that you know when a device is opening and closing, and what parameters are
used to open the device. There are also other technical details such as: you
will need to properly lock down the memory before you read and write to it in
some cases, but not all...if I recall the buffers passed are of the type
METHOD_NEITHER, which requires special attention.
If you need some help along your journey, I am willing to help; as I am sure
others here are. It is not an easy task; but, once you complete it, I am sure
you will be very proud of your achievement.
From: wdmaudiodev-bounce@xxxxxxxxxxxxx
[mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx] On Behalf Of kevinrich47
Sent: Thursday, June 25, 2009 7:05 PM
To: wdmaudiodev@xxxxxxxxxxxxx
Subject: [wdmaudiodev] Re: How to filter microphone voice?
Hi,Could you please tell me which technology in Microsoft is the best choice to
implement a voice changer?Thank you .
----- Original Message -----
From: Frank <mailto:Frank.Yerrace@xxxxxxxxxxxxx> Yerrace
To: wdmaudiodev@xxxxxxxxxxxxx
Sent: Friday, June 26, 2009 9:29 AM
Subject: [wdmaudiodev] Re: How to filter microphone voice?
Note that the only appropriate way to distribute an LFX is as part of a
complete audio driver package. Of course, if this is just for
experimentation/proof of concept then do whatever you want.
Frank Yerrace
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no rights.
From: wdmaudiodev-bounce@xxxxxxxxxxxxx
[mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx] On Behalf Of Tim Roberts
Sent: Thursday, June 25, 2009 5:42 PM
To: wdmaudiodev@xxxxxxxxxxxxx
Subject: [wdmaudiodev] Re: How to filter microphone voice?
kevinrich47 wrote:
Thank you Tim,What I want is not change voice for single app,but for all
application on the machine,
I have consider Virtual Audio Cable,But the price is not affordable for me.
Oh, come on! There's a free version with some features limited for you to try
out, and the full version is only $30. If you have not done audio drivers
before, it will save you MONTHS of frustration.
If you are working on Vista, you can write a Local Effects Audio Processing
Object (LFX APO) to do this. This is a user-mode DLL that lives in the the
audio engine process. There is a sample in the WDK in src\audio\sysfx. It's a
little bit tricky to get one installed to begin with, but there are documents
available to help with that.
--
Tim Roberts, timr@xxxxxxxxx
Providenza & Boekelheide, Inc.
Other related posts:
