Optimal Sound News: Microsoft High Quality Acoustic Echo Cancellation

Microsoft includes High Quality Acoustic Echo Cancellation and Microphone Array Processing in Vista*

For addional information on this subject, click here.

In a departure from previous implementations, Microsoft has included a new High Quality Voice Capture DirectX Media Object (DMO) as part of DirectShow 9. It is capable of performing the following operations:

Acoustic echo cancellation (AEC)
Microphone array processing (Beamforming)
Noise Suppression (NS)
Automatic Gain Control (AGC)
Voice Activity Detection (VAD)

Block Diagram of Acoustic Cancellation and Array Processing DMO

Block diagram of High Quality Voice Capture DirectX Media Object

DirectShow9 is included with the new-for-Vista Windows SDK, which replaces the older Platform SDK. Source code for a sample command-line application using the High Quality Voice Capture DMO is also included in the Windows SDK.

While initially intended for use with USB-based array microphones, guidelines are provided for HD Audio implementations as well. While the Microsoft UAA HD Audio Class driver does not yet support KSPROPERTY_AUDIO_MIC_ARRAY_GEOMETRY, if a 3rd party HD Audio driver supports this property then it should be possible for an application to use the DMO in source mode. An application can choose to use the more-difficult-to use filter mode, which will work with the UAA HD Audio class driver.

For beamforming microphone arrays, only five predefined geometries are supported, two for 2-element arrays and another three for 4-element arrays. Microsoft believes these should cover the most common scenarios for near-term microphone array applications, with more development slated to follow. Additional details are available from the following microphone array documents on the Microsoft web site.

Additional details can be found in the readme file for the demo application. You can get this by downloading the Windows SDK RC1. After you are finished installing it, the readme file can be found on your local hard drive at
C:\Program Files\Microsoft SDKs\Windows\v6.0\Samples\Multimedia\Audio\AECMicArray\readme.rtf

The High Quality Voice Capture DMO is intended to be used by software application developers writing speech applications for Vista PCs, for programs such as Instant Messenger* or Skype*. There is no support for Windows XP, nor for legacy applications that previously relied on the driver to perform this processing. New applications must be written in order to take advantage of this DMO. HD Audio systems with microphone arrays should use a 3rd party driver in order to be compatible with applications which use the DMO in source mode.

Microsoft has stated that the Media Foundation Transform (MFT) will be replacing the DMO as developers transition from DirectX to Media Foundation, so it's possible that AEC and beamforming functionality could show up in the form of a built-in MFT in future Vista updates or follow-on releases.

The block diagram shown above includes analog microphones. Digital microphones can solve many problems inherent in microphone placement on laptop PCs and other applications. For a good overview on using digital microphones for beamforming, check out this article in EEE Times Asia (may require subscription).

Use Optimal Sound services to add an audio expert to your team, and ensure that your systems sound great, are easy to understand, and function properly with Windows Vista. Contact us for more info.

News about HD Audio and Windows* Vista*

Microsoft Releases Updated Guide to Array Microphone Implementation A major update to this white paper from Microsoft provides previously-unavailable details on how the Acoustic Echo Cancellation and Microphone-Array Beamforming processing built into Windows Vista is implemented, as well as how to test it and use it. For more info, click here.

Microsoft Releases New Microphone Array Verification Tool Microsoft has released a command-line tool which can verify whether an audio driver accurately reports microphone-array geometry information and whether the Voice Capture DMO operating in microphone-array mode processes the captured data accurately. For more info, click here.

Microsoft includes built-in Acoustic Echo Cancellation and Microphone Array Processing in Vista As part of DirectShow 9, Microsoft has included Vista-only source code with the new Windows SDK. This source code can be compiled to create a test app which demonstrates the Acoustic Echo Cancellation DMO. While initially intended for use with USB-based array microphones, guidelines are provided for HD Audio implementation as well. For more info, click here.

Microsoft Documents Limit on DTM Test Storage, outlines possible solutions Driver Test Manager (DTM) is the new logo testing infrastructure used for both Windows Vista and Windows XP logo submissions. The default installation is limited to 2 GB of storage, which is enough to get started, but probably not enough for serious DTM users. To find out more options, click here.

Microsoft includes Audio System Effects in Vista Microsoft has released an updated white paper on how to use the Audio System Effects that are included at no charge in Windows Vista. In-box built-in effects include
    Loudness Equalization DSP (LFX)
    Forward Bass Management (LFX)
    Reverse Bass Management (LFX)
    Low Frequency Protection (LFX)
    Speaker Fill (LFX)
    Room Correction (GFX)
    Virtual Surround (LFX)
    Speaker Phantoming (LFX)
    Virtualized Surround Sound over Headphones (LFX)
    Bass Boost (LFX)
For more details, click here.

Microsoft WHQL postpones date that systems must pass audio fidelity tests On Thursday, August 30, Microsoft's Hardware newsletter announced that systems will not be required to pass the audio fidelity portion of the logo tests until June 1st, 2007. However, all submissions must include complete logs from the audio fidelity tests, which requires an Audio Precision System 2 test set to be connected to the DTM test environment. For more info, click here.

Audio Fidelity Testing Now Required for Vista Audio Device Logo Microsoft has released Windows Logo Program (WLP) specification 3.07, which has significant impact on the audio development and logo submission process for Windows Vista. Audio devices and drivers are now required to included passing audio fidelity test logs in order to qualify for device logo. For more info, click here.

Windows Vista RC2 to be released October 6, 2006 Recently disclosed details of Windows Vista RC2 release indicate that Vista build 5743 will be released Friday, October 56, 2006, and will be designated as Vista Release Candidate 2 (RC2). For more info, click here.

Audio Design Line Publishes excerpt from HD Audio Book EE Times affiliate Audio Design Line, has reprinted Chapter 9 of the new book from Intel Press, High Definition Audio for the Digital Home: Proven Techniques for Getting It Right the First Time by David Roach, Scott Janus, and Wayne Jones. This chapter of the book describes the complete signal audio signal path of a modern PC, including both hardware and software audio components. For more info, click here.

Microsoft publishes sample code for 3 different Audio System Effects Microsoft has released sample code for three different sample Audio System Effects along with a retitled document describing how to use them. For more info, click here.

Microsoft Multimedia VP Amir Majidimehr writes about Audio System Effects included in Windows Vista Read first-hand the thinking behind the new Audio System Effects included in Windows Vista. For more info, click here.

Microsoft Channel 9 Video on audio hardware for Windows Vista features Hakon Strande and David Roach Check out the 38-minute Channel 9 video featuring Hakon Strande of Microsoft and David Roach of Optimal Sound discussing the audio hardware in Vista. For more info, click here.

Microsoft Releases HD Audio Pin Configuration Test Guidelines Microsoft has released a paper entitled Microsoft HD Audio Pin Configuration Implementation Test Procedure Guidelines. The paper describes how to use a Microsoft-supplied pin configuration utility to generate and validate a set of verb tables which is embedded in the system BIOS. The verb table describes the audio configuration of the system. If the verb table is configured incorrectly, the UAA HD Audio class driver will not function properly and the Windows Vista logo tests will fail. For more info, click here.

News about HD Audio & Vista

Microsoft includes High Quality Acoustic Echo Cancellation and Microphone Array Processing in Vista*

News about HD Audio and Windows* Vista*