Jump to content
Home

News

CLAAUDIA launches new AI transcription solution for researchers

Published online: 21.11.2023

This summer, researchers from AAU gained a new and powerful ally in their research work. Thanks to CLAAUDIA's latest release, Whisper Transcription, researchers can now transcribe video and audio files using artificial intelligence and no longer spend hours transcribing data.

News

CLAAUDIA launches new AI transcription solution for researchers

Published online: 21.11.2023

This summer, researchers from AAU gained a new and powerful ally in their research work. Thanks to CLAAUDIA's latest release, Whisper Transcription, researchers can now transcribe video and audio files using artificial intelligence and no longer spend hours transcribing data.

Textt: Nana Møller Larsen, ITS     Photo: AAU

Behind Whisper Transcription is a dedicated team from CLAAUDIA consisting of Data Steward Freya Vamberg Delfs and Data Scientists Robert Smith and Pelle Rosenbeck Gøeg, the latter being responsible for the development of the application. Freya explains:

- The idea for Whisper Transcription came about because we received a lot of enquiries from researchers seeking guidance on which transcription services, they could use for their data processing. The problem is, however, that we can't guarantee secure processing of sensitive data with the different transcription services we've been asked about," says Freya and continues:

- That's why we at CLAAUDIA started developing our own solution that can process sensitive data and which we can provide support if researchers need help.

Secure and fast transcription

CLAAUDIA's new transcription tool is a valuable resource for researchers working with qualitative data such as interviews, as it often contains confidential or sensitive information and therefore needs to be handled and stored securely. With Whisper Transcription, researchers can get help transcribing their data without compromising security. The application is available in version 1.0 on DeiC's interactive HPC platform, UCloud, which provides a secure infrastructure that allows researchers to safely upload and store both confidential and sensitive data on the platform. In fact, the application is the only AI-based transcription solution for Danish researchers that guarantees that their data is stored securely and locally on Danish servers at DeiC.

- Whisper Transcription is based on OpenAi's Whisper language model and the application allows researchers to upload and transcribe individual files as well as entire folders of video or audio files. The application can transcribe in real-time - sometimes even faster depending on the CPU you choose to utilise, Freya explains.

With Whisper Transcription, users can choose up to seven different output formats, including the option to export their transcribed data in plain text format or as an srt file, which is the format used for subtitles. In addition, it is possible to transcribe your data in several different languages, so even if English is spoken in your audio file, you can choose to have the application transcribe in Danish.

A tool for researchers, students and employees

Although CLAAUDIA has developed Whisper Transcription based on a need from researchers at AAU, researchers from other universities, students and employees will also be able to benefit from the new tool. The application is freely available on UCloud for anyone with a wayf login, says Freya, who also sees an advantage in that the new transcription solution can help more researchers get acquainted with UCloud:

- We hope that Whisper Transcription will introduce researchers, who mainly work with qualitative data, to UCloud. In doing so, they will hopefully discover that, in addition to Whisper Transcription, there are a lot of valuable resources that they can use for data processing in their research," says Freya.

In the CLAAUDIA theme, they are currently investigating the possibility of introducing new features in future versions of the application:

- We are working on making it possible for users to select Word file as output format. In addition, speech recognition, which will enable the application to recognise the different speakers in a given audio file, is in our development pipeline. We hope to present the aforementioned features in the next update, says Freya.

In the months that the first version of Whisper Transcription has been available, CLAAUDIA has already received a lot of positive feedback from researchers who have used the application, which is the eighth most used application on UCloud.

Since the article was written and published, CLAAUDIA has updated Whisper Transcription to version 1.1. In the new release, additional export options have been added, including the ability to export transcribed data as an MS Word file. Additionally, the output can be exported and compressed in .zip file format, and to increase security, users can password protect .zip files using AES encryption, allowing users to transfer the files securely and confidentially. In addition, the new version allows file names to contain spaces without preventing the user from uploading them and using them with the application.

Webinar: Getting started with Whisper Transcription

ReachAAUt Researcher Network is hosting a webinar where you can learn more about Whisper Transcription, how to access the application, and how it can help you as a researcher.

The webinar will be held on November 30th at 09-10.

If you are interested in attending, you can find the invitation here. You can also contact Dagmar Knudsen Fallesen at dagmarkf@its.aau.dk.

Useful links

If you want to use Whisper Transcription for your data processing, you can log in and find the application on UCloud here.

CLAAUDIA has made a demonstration video showing the different features of Whisper Transcription, you can watch it here.

If you are curious to read more about OpenAI's Whisper language model, which Whisper Transcription is built on, click here.

Enquiries to CLAAUDIA are welcome at claaudia@aau.dk.