Speech can facilitate communication in many situations. It’s no wonder that voice input on smartphones and assistants such as Alexa, Siri and others are booming. Industrial applications can also benefit from this, say Dr. Diane Hirschfeld and Martin Wenzel from voice Inter connect in an interview.
Ms. Dr. Hirschfeld, Mr. Wenzel, speech in professional communication systems – what are the challenges here?
Dr. Diane Hirschfeld: Conference calls from home with a PC work very well – as long as we are in a quiet room. But almost everyone has experienced what it is like when a drill in the apartment next door is louder than your own voice, or when you can hear your conversation partner`s laptop fan instead of clear speech. This is annoying when working remotely, but carries no risks. However, this is completely different in the case of emergency call boxes, on train platforms, in special vehicles, industrial plants or control rooms. Here, speech must be transmitted reliably and crystal clear, even in the presence of loud ambient noise. This places high demands on the design of communication devices and signal processing. Additionally, data protection issues and security play a major role in those applications.
Where do we currently stand and what new developments can we expect in intercom technology?
Dr. Hirschfeld: At embedded world we will be showing modules from our vicCOM family that can be quickly and easily integrated into individual devices and expand them with functions for hands-free or intercom communication. They are available for a wide range of ransmission technologies – from two-wire bus to IP and LTE. Thanks to noise suppression and sophisticated filters and algorithms, they ensure the best possible voice quality and intelligibility – even when quite literally »you can’t understand a single word you are saying« because of the background noise. Our algorithms automatically detect where exactly in a room a speaker is located and direct their focus towards that position, even if the speaker is quieter than the surrounding sounds. The algorithms can distinguish speech from background noise, filter the signal and thus enable reliable communication even under difficult conditions.
And what about voice control – is there already an »Alexa« for technical devices?
Martin Wenzel: As far as voice control is concerned, our hardware voice control modules and our software voice control solutions are in no way inferior to online tools such as Alexa. At the same time, however, they work completely offline without requiring an online connection during operation and can thus also be certified for industrial and safety-critical applications. In this way, professional applications also benefit from simple, convenient and intuitive operation. With their Natural Language Understanding/Processing unit and the latest AI-based semantic analysis, they evaluate the meaning and context of the recognized words and translate them into the appropriate control signals. Status information can also be retrieved by voice, because text-to-speech can be used to automatically generate voice outputs from database records or status information. The audio prompts do not need to be recorded or updated for this. The voice control solution »vicControl industrial« is now available in more than 30 languages.
Which are typical applications for intercom and voice control?
Martin Wenzel: Facility manufacturers integrate our OEM modules into their products for building communications; use them for handsfree devices or for voice control of medical technology and industrial applications. For this purpose, they use our standard modules, which are suitable for many applications and can be integrated quickly and easily. Or we develop individual hardware modules that are adapted to the respective requirements and thus are cost-efficient for series production. Often, our expertise in intercom, voice control and signal processing complement each other perfectly, for example in infotainment devices for commercial vehicles – with noise-robust handsfree telephony and integrated voice control for status queries and the control of attachments while driving.
How does voice Inter connect support its customers during integration?
Dr. Hirschfeld: We provide individual support for each project: from the selection of an appropriate standard module to the development of new or customized electronics. In this context, hardware development is only one aspect of our offering: at voice Inter connect we have an interdisciplinary team with specialists in hardware as well as software development, signal and speech processing. These experts support customer-specific product tevelopments in topics such as AI, acoustics, ergonomics and application development. Hereby we offer our customers demandoriented support from conception to fast and cost-effective implementation of applications that improve the dialog between people and between man and machine.
voice Inter connect’s intercom module series »vicCOM« offers full-duplex communication with clear voice quality and can be quickly and easily integrated into individual communication devices, thanks to its many connections including web interfaces. For the first time, voice Inter connect will be showing the new module »vicCOM GSM 2« for IP and 2-wire bus data transmission at embedded world. It supports LTE and will also be available in a version for rail applications in accordance with the GSM-R standard. Compared to ist predecessor, vicCOM GSM 2 has two slots for SIM cards and four directly connected call buttons. These can also wake the module from sleep mode and, in conjunction with the charge controller, make possible extremely power-saving and self-sufficient intercom systems.
Also on display in a new version will be the embedded voice control »vicControl«, which is available in over 30 languages – from Benelux languages to Norwegian and Swedish plus newly added languages like Arabic Gulf and Sichuanese Chinese. The voice control is easy to implement and setup, it uses Natural Language Understanding (NLU) and AI-based semantic analysis. A recent innovation is the output of status messages via text-to-speech. This allows any text to be read out without having to pre-record audio prompts or update them in the future. Thanks to local voice processing, vicCONTROL industrial does not require an Internet connection and therefore meets the highest standards of security and data protection.
Until now, the biggest challenge for speech user interfaces was reliable speaker recognition in the presence of loud background noise. This is exactly what voice Inter connect’s toolbox solves with its extensive library of algorithms for speech detection, beamforming and noise suppression. It was developed for high performance and reliability and in its latest version, offers the automatic spatial localization of a speaker. The associated evaluation kit »vicDiva« is available now. voice Inter connect also offers comprehensive and platform-independent services for the design and development of individual intercom applications, voice control and signal processing products. These include acoustic measurement and parameterization services as well as support in design and component selection up to application development for the control of machines and systems.