Voice Controlled Audio Devices
Any smart device, any measurement
SoundCheck offers simple, fast and accurate ‘open loop’ testing of any smart device – no matter what the form factor, functionality, connector, or additional features.
SoundCheck offers simple, fast and accurate ‘open loop’ testing of any smart device – no matter what the form factor, functionality, connector, or additional features. It measures a full range of characteristics such as driver performance, microphone array performance, speech recognition and voice quality metrics via wired, wireless/Bluetooth, cloud, USB, or debug port connections.
SoundCheck’s long history as a multipurpose electroacoustic test system means that Listen, Inc. has unrivalled expertise in the various tests used in voice activated audio devices – e.g. microphone array testing and beamforming, driver testing, speakerphone testing to TIA standards, active noise cancellation, voice quality measurements and speech recognition metrics.
Furthermore, Listen, Inc. has been making ‘open loop’ measurements – measurements on devices with no direct access to the microphone or speaker for longer than any other test equipment supplier. This has resulted in extremely sophisticated
algorithms for optimizing the speed and accuracy of open loop tests, as well as the expertise to guide you through the process.
Smart Speaker / Smart Device Measurements Include
- Traditional measurements such as frequency response, directionality, distortion
- Telephony tests to TIA standards
- In-situ measurements such as voice recognition in noisy environments
- Signal to noise with background noise simulation
Directionality of smart speaker microphone
The SoundCheck System
SoundCheck’s modular combination of hardware and software is cost-effective, flexible and expandable.
At the heart of the system is the SoundCheck software. Powerful, fast and accurate, it measures every audio parameter from the R&D laboratory to the production line. With smart triggering and sophisticated post-processing algorithms to allow asynchronous measurement (commonly known as ‘open loop testing’), SoundCheck offers incredible flexibility for testing using virtually any interface including Bluetooth, cloud-based systems, A2B® bus automotive interfaces and more. Complete flexibility in stimulus choice enables the use of real and complex test signals such as multitone, speech, and even music, to test devices.
Repeatable, automated tests are quickly and easily created, modified and saved using the simple point-and-click interface. Several easy to modify sample sequences for measuring voice-activated devices are provided. These include sequences for open loop microphone and speaker test, as well as more complex sequences for specific test standards such as TIA 910 and background noise simulation to the ETSI ES 202 396-1 standard.
The software also controls the audio measurement hardware – an audio interface, measurement microphone, power supplies, and amplifiers and speakers for background noise generation.
Smart Speaker test results showing frequency response and distortion
Listen’s smart device test hardware includes:
AmpConnect™: Listen’s all-in-one hardware includes an audio interface, amplifier, power and gain for up to 2 SCM or IEPE microphones, digital I/O, and
integrated impedance circuit for all voice-controlled device tests.
AudioConnect™: A compact audio interface with power supply and gain for up to 2 SCM™ microphones. This is a very cost effective approach for basic smart device testing.
SCM™ Measurement Microphones: SCM measurement microphones are robust, accurate, low noise, stable and cost-effective.
SoundConnect 2™: A 2-channel microphone power supply (SCM, IEPE, 200V polarization) ideal for applications where polarization, TEDS, and high and low pass filtering are required. It also provides autoranging when used with SoundCheck to maximize the dynamic range of a measurement.
Bluetooth Audio, MEMS, A2B and other Interfaces: Bluetooth speakers, headphones and car audio head units, connect to SoundCheck via Bluetooth
interfaces, offering full control over all Bluetooth protocol settings and explicit control over the CODEC choice and transmitter power. Mentor’s A2B
interface can also be connected to, and controlled by SoundCheck for automotive testing applications. Other multichannel audio interfaces, MEMS
interfaces, amplifiers, current monitors, and 3rd party hardware such as head and torso simulators are available for more complex test requirements.
R&D Smart Device Testing
In the R&D lab, SoundCheck measures everything from simple driver response to microphone beamforming, speakerphone performance, voice recognition, voice quality and more. SoundCheck’s sequence editor makes it simple to re-run tests and compare results as designs are modified. Powerful test capabilities, with many stimuli, analysis, post-processing and statistics options allow you to make all the measurements you need, plus offer unrivalled flexibility to calculate any parameters arising from these without leaving the system.
Production Line Smart Device Testing
On the production line, SoundCheck offers extremely fast and accurate testing of both finished goods and subcomponents such as speaker or microphone array sub-assemblies. It is simple to operate, offers high immunity to background noise, and can carry out complete production tests including frequency response, our advanced Perceptual Rub & Buzz, THD, loose particles, polarity and phase. Results can be compared to pre-set limits or a reference standard, and can be presented as a simple audible or visual pass/fail indication, a detailed failure mode analysis, or automatically written to a database. SoundCheck interfaces with barcode systems, footswitches and PLCs, and can be controlled via 3rd party programs such as National Instruments Test Stand, LabVIEW®, Python, C#, etc. for full integration with automatic production lines and large-scale test systems.
Smart Speaker Test Example
With both sound and listening quality being differentiators between competitive models, robust audio testing of both the speaker and the microphone array is essential.
However, these measurements present two significant challenges:
- There is typically no direct path to inject or extract response signals from the embedded speakers and microphones; rather than playing back and recording signals to local storage media, they store and playback signals from Internet based cloud services.
- The use of voice recognition for control makes activating these devices challenging. These characteristics necessitate various additional steps to be taken in the testing process to ensure accurate and meaningful measurements.
With some specialized algorithms, SoundCheck is uniquely positioned to measure smart speaker parameters. Additional specialized equipment is not required, but it is important to establish the best method of accessing recorded wave files, and to correctly configure the test to take into account wake words, recording time, latency and sampling rate errors.
Watch our short 4 minute video for a demonstration of how we can measure both embedded speaker and microphone array parameters such as frequency response and distortion.
Our white paper ‘Smarter Measurement for Smarter Speakers’ outlines how we established a method for accurately testing one of the world’s best-selling smart speakers.