MEMS microphones drive next-generation voice codec

MEMS microphones drive next-generation voice codec

Feature articles |
By Peter Clarke


Micro Electrical Mechanical System (MEMS) technology has found its sweet spot in the IoT, enabling a wide range of tiny, low-power sensors which are transforming the way in which we interact with our devices. Thanks to this innovative technology people are spending less time tapping on keyboards, moving pointing devices, or operating switches, as digital audio-visual communication tools become more common-place and reliable.

MEMS microphones, in particular, offer a wide spectrum of possibilities for voice-based applications, including intruder alarms, smart speakers, in-car control systems and many more. With the cost and time to market pressures in the IoT space, the ready availability of high-quality digitisation technology is critical to the development of voice-based devices. However, with the market’s recent focus firmly on multi-media, voice technology has been neglected, resulting in a dearth of suitable interface devices for MEMS microphones.

In this article we look at how this situation is changing as developers seek to capitalise on the growing opportunities presented by MEMS microphone technology.

MEMS technology

MEMS is a term covering the microfabrication processes involved in realising moving mechanical parts at a microscopic scale. By integrating both electrical and mechanical components onto a single chip, MEMS technology is used to produce microscopic, electro-mechanical sensors with sizes ranging from a few microns up to millimetres.

MEMS sensors are easily mass produced at much improved economies of scale and their compact size, along with vastly reduced power consumption and higher sensitivity, when compared with traditional mechanical versions, have led to their widespread adoption, particularly within the Internet of Things, (IoT). Nowadays we all carry a variety of MEMS devices around in our smartphones, smart watches, and fitness trackers. A MEMS gyroscope, for example, in a smartphone, weighs less than a milligram and is no bigger than a grain of sand.

Electrostatic transduction, involving a variable capacitance, is traditionally the most popular type of MEMS sensor, since micromachined silicon can be doped to provide conductivity and no additional material is required during fabrication. Figure 1, for example, shows the block diagram for a digital MEMS microphone. The transducer element contains a moveable membrane and a fixed backplate, with the capacitance between them changing as the air pressure from sound waves causes the membrane to flex. The onboard amplifier converts this capacitance into an electrical signal, which is then converted into binary digital format by the ADC.

Figure 1: MEMS microphone block diagram (Source:

Although MEMS microphones are available with both analogue and digital outputs, it is usually more straightforward and convenient to design and implement solutions with a digital data interface, leading to an increase in the popularity of digital MEMS microphones. This is particularly true when operating in hostile or noisy electromagnetic environments. Digital MEMS microphones use pulse density modulation (PDM), which is relatively immune to noise and crosstalk, to produce a highly oversampled single-bit data stream as output.  PDM has the added advantage of enabling two digital microphones to share a common clock and data line, with each microphone configured to generate its output on a different edge of the clock signal.

Digital MEMS microphones offer high SNR, low power consumption, good sensitivity, have excellent temperature characteristics and are available in very small packages that are fully compatible with surface mount assembly processes. 

With more and more modern smartphones and other smart devices able to process language and voice commands, the demand for high quality, compact sound technology, including microphones is showing strong growth. Along with smartphones, MEMS microphones are used in earphones, laptops, and hearing aids as well as a growing range of IoT devices, such as wearables, remote control equipment and home appliances such as refrigerators, air conditioners and service robots.

A microphone on its own is of limited value in a smart device, ultimately the output signal must undergo further processing before it can be analysed by the system controller and therefore application developers must implement a suitable interface.

The next-generation voice codec

Interfacing to a microphone has long been the role of the voice codec however traditional codecs were designed to be compatible with electret microphones rather than with the modern digital MEMS version. This previous generation of voice codecs lack a number of the functions required to interface with a MEMS microphone, including decimation filtering, which is required to convert PDM into the PCM format required for further processing. Also, with many applications including a loudspeaker as well as a microphone array, a Class D amplifier, with its superior efficiency, has become an essential part of any MEMS microphone interface.

Recent industry focus on codec development prioritised multimedia applications, driven by the rise in smart phones and tablets and these devices have not brought solutions to the specific requirements of voice applications. This situation has led to a gap in the market for up-to-date voice codecs, potentially placing the application developer in the undesirable position of having to add extra circuitry to existing devices.

Fortunately, a number of device manufacturers have recognised the need for a next-generation voice codec, integrating enhanced functionality into small, low-cost devices with levels of power-efficiency which are compatible with battery powered applications.

An example of this new generation of voice codecs is the CMX655D from CML Microcircuits shown in figure 2.

An ultra-low power voice codec, CMX655D is specifically designed to support the latest MEMS microphone technology for always-on digital voice applications. With its ultra-low power consumption, drawing typically 500uA in stereo record mode and less in low powered  listening mode, the CMX655D is ideal for battery-powered devices. The chip supports the bandwidths commonly used for both conventional telephony (300Hz to 3.4kHz) and HD voice (50Hz to 7kHz), as well as a full audio band 50Hz – 20kHz mode.

In addition, the device supports external noise cancellation applications by interfacing simultaneously with two microphones and keeping phase matching between the two paths identical throughout the device. The CMX655D package also includes an integrated high-efficiency Class D speaker driver, providing an output of up to 1W. This functionality is generally not available with other devices, including general purpose DSPs, and so would typically require an additional external IC.

Figure 2: Simplified block diagram of the CMX655D voice codec

By selecting a device such as the CM655D, the developer simplifies the hardware design task and achieves a fully functional MEMS microphone system ready to be integrated into the specific application. Although device selection is therefore critical to the reduction of development cost and time to market, it is not the only factor to be considered when choosing a solution. Development environments are crucial tools when attempting to integrate any device and, in the competitive IoT market, the choice of device is heavily influenced by the functionality, ease of use and economics of this supporting toolkit.

Open-source development toolkits

Many manufacturers offer bespoke hardware and software development kits to support their devices however, with reduction in development time and cost being critical in the IoT world, developers are increasingly turning to open-source options. In this context systems such as the Raspberry Pi (RPi) have moved out of the hobbyist and educational realms to become practical tools for the professional engineer.

The RPi Model B+, released in 2014, introduced the Hardware Attached on Top, (HAT), interface, making it easier for third-party hardware to plug directly into the RPi infrastructure (Figure 3). This powerful capability, along with the RPi’s Linux-based operating system, enables rapid development, testing and prototyping, helping design teams get to the proof-of-concept stage much more quickly. Most importantly, this approach significantly lowers costs compared to the use of bespoke development boards and proprietary software and also provides a solid foundation for further development.

Figure 3: The EV6550DHAT add-on boards from CML

With the release of the CMX655D, CML Microcircuits have chosen to follow this open-source approach. By creating the EV6550DHAT evaluation board for the CMX655D codec solution in HAT form, CML have made their new device easily available to the Raspberry Pi community.

Furthermore, recognising that it is not always practical for a developer to spend time acquiring the knowledge required to successfully integrate the audio functionality into the application, CML have also made a Linux driver available for the CMX655D-RPi configuration. This-low level driver enables use of the development environment with any audio application that follows the standard Advanced Linux Sound Architecture (ALSA) interface. The ALSA architecture uses a very flexible interface, providing a level of abstraction which greatly simplifies the development of applications which use sound.


MEMS technology is finding multiple applications within the IoT space, enabling mass produced, low-costs sensors for a wide range of scenarios. Digital MEMS microphones are one example of how MEMS technology is enabling new ways of human-device interaction, enabling high quality audio in applications such as smartphones, smart speakers, wearables, and many others.

A new generation of codecs supports the integration of MEMS microphones, integrating the required interface functionality into low-cost, low-power devices with small form factors. By enabling their devices to function with open-source development tools such as the RPi and the ALSA software framework, innovative manufacturers such as CML Microcircuits are helping their customers to reduce development costs and speed up time to market.

David Brooke Is the wireless voice and data product manager for CML Microcircuits Ltd.

Related links and articles:

News articles:

Secure 2.4GHz radio transceiver has 12km range

Adaptive MEMS microphone stays asleep longer to save power

MEMS market decline in 2020 includes options for long-term growth

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles