Beey also excels in Norwegian thanks to the EEA and Norway Grants

November 8, 2022 in News

Beey also excels in Norwegian thanks to the EEA and Norway Grants

Are you curious about the development and addition of a new language to Beey? On 20 September NEWTON Technologies attended a meeting about our implementation of the EEA and Norway Grants. We introduced the NORDTRANS project, in which along with the Technical University of Liberec we are developing technology for automatic speech-to-text transcription for Norwegian (and eventually Swedish).

Why Norwegian?

Developing Germanic languages

As a Czech company, we primarily dealt with Czech and other Slavic languages. Germanic languages were the next logical step in the development of recognition software for the European market.

Collaboration with the unversity

The Technical University of Liberec, where the technological background of the recognition software is being developed, has previously collaborated with the Norwegian University in Trondheim (NTNU) on other projects, so they have retained the contacts and established good relations needed to acquire and navigate new data.

TACR Interview

The Technology Agency of the Czech Republic, which mediates funding from the Norway Grants to specific projects, has approached us before and organized an interview with our linguist Lenka Weingartová as part of the Meltingpot forum at the Colours of Ostrava festival.

Zástupkyně z TA ČR na Setkání k iimplementaci norských fondů

The Recognition Software

For developing the Norwegian model and all other upcoming languages, we've been using a new generation of speech recognition technology. The so-called end-to-end recognition is based on a single massive neural network. This artificial intelligence is trained for the desired language using machine learning algorithms, which primarily requires a vast amount of data: audio recordings and their verbatim transcripts. Given enough quality data, this version of the recognition software is very accurate, typically achieving accuracy of over 90% for clear recordings, and requires significantly less manual maintenance and updates. The disadvantage may lie in getting the data of the required quality, or relatively limited possibility to influence the output from the already finished model.

Potential use cases

Such high quality speech recognition device has many various uses, especially when implemented in our Beey app. The automatic speech-to-text transcription can be edited further to achieve 100% accuracy. The result can be automatically or manually separated by speakers, structured by topics and so on. Beey also maintains the time stamps connecting each word to the audio track, which allows the user to watch or control the playback by the caret position in the transcribed text. However, the main benefit shows in creating subtitles, which can be generated automatically in one click. There is also the Subtitle mode tool, which includes advanced subtitle revision, professional editing and subtitle previews both in the video and in the audio waveform visualization.

Other areas of speech-to-text transcription application include:
  • TV, radio and internet broadcast online monitoring
  • transcript and documentation of parliamentary meetings and other administrative offices
  • transcription of council meetings, lectures, conferences etc.
  • fulltext search in audio archives
  • improving SEO when publishing a video or podcast with a transcript
  • captioning videos for people with hearing disabilities
  • instant subtitling of foreign language videos using the Beey Translate automatic translation feature
Alternative Text

Presenting the NORDTRANS project

We are proud to have been selected from twenty other projects in the Research category to present the Norwegian recognition module as an example of good practice in the use of Norway Grants. In addition, we were also pleased that the Ambassador of the Kingdom of Norway and other representatives of the Norwegian Embassy in Prague attended the meeting to witness the quality of the Norwegian transcription during a live demonstration in the Beey app.

Try Beey for free

KAPPA programme financed by EEA Grants

Within the framework of cooperation with TUL, NEWTON Technologies is involved in the KAPPA Programme for Support of Applied Research, Experimental Development and Innovation, mediated by the Technology Agency of the Czech Republic. Our joint project was supported as one of 25 projects out of a total of 163 submitted proposals.

more about the kappa programme
Alternative Text