Kymberly Christman, Maxim

James Wihardja, Conexant

Doug Peeler, Dell

Goal
Create a new form of acoustic monitoring for the PC space that can be used to improve user experience with minimal user interaction. 

Problems to Solve

1. Ensure system is always ‘appropriately’ loud given various noise environments.
1. If quiet ambient, SPL is reduced and power is saved.
2. Use capture (acoustic feedback) to dynamically improve audio quality at the ear.
3. Ensure perceived loudness is constant with varying content.
4. Limit SPL based on IP address/location to enable group policy of noise pollution.
5. Enable optional automated speaker EQ.
6. Enable in-air speaker performance testing.
7. Works with multiple devices.

Potential Outcomes

1. Automatic volume control
2. Power Savings
3. Adaptive Speaker Optimization
4. PC “Stealth mode” (reducing noise pollution from system)
5. Corrective Hearing
6. In-air Self Test

Steps to Implementation
Smart Ambient Sound Sensors will be introduced in multiple phases.

1. Calibrated digital mic is the sensor.
2. Capture limited to systems with captive speakers and sensor.
3. Define new data type which includes frequency, amplitude, etc.  
4. Data is provided in three types:
1. Event Driven
2. Polled
3. Real time Stream
5. No phase data included so data cannot be reconstructed and privacy rights are protected.
6. Stream expanded by meta data tags.
1. Room acoustic level in absolute dB.
2. Magnitude Spectrum.
7. Comprehend dynamic devices.

The Problem
Volume control in modern PCs is unpredictable and disruptive.  Contributors include varying content levels, multiple volume controls per application and device, non-standardized controls and UI, and unpredictable room and environment characteristics.   All lead to a poor user experience.

In addition, modern notebook PCs suffer from the perception and often the reality of poor audio quality. The group identified a number of issues such as speaker physical limitations, poor design practices at many PC OEMs/ODMs, and a lack of a standardized “in air” testing standards.

Each audio usage model has implicit expectations which are typically not met on a laptop platform.  PCs have not kept up with ‘quality experience’ precedents set by consumer electronic devices, including performance, real-time communications, and content playback. 

Solution
The group identified the use of a using a calibrated digital microphone – which the group coined a “Smart Ambient Sound Sensor” – to improve the quality of the user experience. The primary focus for use of this Ambient Sound Sensor was quickly narrowed to the PC market where the characteristics of the speakers and approximate head position can be fairly established. The Ambient Sound Sensor might be an existing piece of audio infrastructure, or might be a separate sensor specifically designed for this purpose.

The group envisioned a two stage approach.  Stage One would be primarily used by components in the existing audio ecosystem (HDA driver, APOs, & existing digital microphones).  The Ambient Sound Sensor would monitor the audio environment either in real time or periodically, and send an audio stream with information on crucial audio characteristics of the audio environment (loudness levels, active frequency bands, etc) with potential for additional metadata. The standards for this audio stream would be driven by the HD Audio working group with input from key stakeholders such as IHVs, audio infrastructure vendors, and APO stakeholders. Issues identified to be solved by the stage one solution include volume control depending on room noise conditions, inverse filtering based on speaker characteristics, and more advanced IHV tools for testing and configuring systems before shipment.

Stage Two envisions making the Ambient Audio Sensor Stream available to the system. The group identified a number of steps needed to be taken, such as defining an Ambient Sensor Class, which will have different characteristics then today’s audio endpoints. By making the stream available to the system we envision a number of additional problems that could be addressed, such as home stereo tuning through the innovation of third party software developers.
 
Concerns
Doesn’t break existing speaker tuning investments.

Privacy of captured audio streams can be protected by converting captured data to a form which cannot be reconstructed into a listenable audio stream. This can be accomplished in many ways, including converting into frequency domain information and possibly throwing out the phase information. This conversion can take place at the following points in the system, shown in descending order of security:
            Microphone output
            Codec
            Chipset
            Input LFX

Government agencies may require capture to be permanently disabled.

Cost of implementation and offsetting ROI.

How to measure success in business case?

Action Items

section 7

next section

select a section:
1. Introduction  2. Speakers  3. Executive Summary  
4. Wrap It Up -- Creating a New Middleware Marketplace for Digital Entertainment Platforms
5. The EduMusiTainers™
6. So you want to work in game audio?
7. Smart Ambient Sound Sensor
8. New Creators and New Creative Tools – Understanding the New Ways to Make Music
9. Working as a Creative Professional in a Corporate Environment
10. "PRAGMA" Rebooted (pet rocks and game music alliance) A Project Bar-B-Q 2008 rogue group updating a Project Bar-B-Q 2004 rogue group
11. Schedule & Sponsors