AI-generated
Neural voices are conquering the world of public transport – now with ad-hoc announcements too
Interest in our text-to-speech solution with neural voices has continued to gather momentum over the past two years. Numerous projects with transport operators have been implemented successfully. The use of neural voices has now become even more attractive thanks to a new function for ad-hoc announcements directly from the control centre.

Rethinking passenger information
The success of the text-to-speech solution has been particularly evident in practice wherever passenger information needs to be available quickly, clearly and without complex production processes. The question today is therefore less whether neural voices offer added value, but rather how extensively transport operators can use them for modern passenger communications.
In the event of short-notice timetable changes, diversions, construction works, events or exceptional operational situations, information often needs to reach passengers within the shortest possible time. With neural voices, announcements can be generated directly, listened to and provided in high speech quality – without a recording studio, lengthy coordination processes or additional production effort.
This means that the days when new announcements first had to be commissioned, recorded and post-processed are increasingly becoming a thing of the past. Stop, line and destination announcements, as well as distance- and time-controlled announcements, can now be created with just a few clicks and updated flexibly.
Ad-hoc announcements directly from the control centre
A new area of focus is their use in the control centre: employees can create ad-hoc announcements within seconds and transmit them immediately to affected vehicles, lines or stops. This ensures that up-to-date information reaches exactly the passengers or precisely those drivers who need it – for example in the event of disruptions, line closures, replacement services, or safety and service notices.
Natural-sounding announcements with a system behind them
The neural voices sound natural and professional. In many cases, no further phonetic post-processing is required, as even complex terms are reproduced reliably and clearly. The basis for these capabilities is provided by new algorithms based on AI deep learning.
In addition, various voices and languages are available. This means that not only operationally relevant passenger information, but also service and marketing texts can be played out in a targeted way when required – offering added value for both passengers and transport operators.
Customers ask for ad-hoc announcements
The request for a way to make ad-hoc announcements from the control centre came to us several times from customers, including from Braunschweig.
At the start of the test phase for the text-to-speech solution, around a year and a half ago, we received the following feedback from Carsten Borchers, who is responsible for service planning and operations control systems at BSVG Braunschweiger Verkehrs-GmbH:
«We are currently testing this function and have already received positive feedback on it from our driving staff. Our colleagues in the control centre would be even more pleased if they also had the option to ‘record’ something immediately. We very much hope that this option can be implemented.»
Relief for the control centre, added value for passengers
ebblo has since successfully implemented this function with ad-hoc announcements. We are delighted to be able to support our customers even better in their day-to-day operations with these new technical possibilities.
Carsten Borchers is also enthusiastic about the ad-hoc announcements:
«It is very good to see that ebblo responded quickly to our request and implemented ad-hoc announcements. For the control centre, this will provide significant relief, and for passengers it will mean a clear improvement in information.
The current approach of putting together passenger information from voice building blocks will therefore no longer be needed. At present, we still painstakingly assemble individual pre-recorded snippets created in LIO-Data into a sentence. This is very time-consuming and often does not fit the situation perfectly.
With the neural voice, we can now achieve a much better result for our passengers ‘all in one piece’.»
Get to know our voices
Would you also like to keep your announcements up to date at all times, provide ad-hoc announcements directly from the control centre, or find out more about the use of neural voices?
We would be happy to inform you about the possibilities and requirements. Please contact your usual contact person directly or send an email to [email protected].