Microsoft Garage Introduces Group Transcribe: A Collaborative Transcription Solution

A novel project originating from Microsoft’s internal incubator, Microsoft Garage, presents a unique approach to meeting transcriptions. While numerous real-time transcription applications are currently available for smartphones – such as Otter.ai and Google’s Recorder app for Pixel devices – Microsoft’s Group Transcribe app re-envisions meeting transcriptions as a collaborative undertaking.

The core concept involves each attendee recording the meeting independently on their own device, aiming for enhanced accuracy. Furthermore, the application facilitates real-time translation across more than 80 different languages and locales.

Initiating and Joining a Session

To begin, one participant initiates the meeting on their device. Other attendees can then be invited to join the session through Bluetooth connectivity, a scannable QR code, or by sharing a unique link.

Once participants have joined and the meeting commences, a live transcript will be displayed on each individual’s device.

Leveraging AI for Accuracy and Speaker Attribution

Powered by advanced AI speech and language technologies, the app achieves higher transcription accuracy. Speaker attribution is also improved by analyzing the audio captured by each phone’s microphone.

The system determines the device closest to each speaker by comparing voice volume levels. This allows for accurate labeling of speakers within the transcript, a common challenge for single-device transcription apps.

Real-Time Translation Capabilities

Participants who prefer to speak in their native language can utilize the app’s translation feature. The app will then deliver translations to other attendees’ devices in their respective languages.

Accessibility and Inclusivity

Microsoft designed the app with accessibility as a key consideration. It aims to enhance participation for individuals who are deaf, hard of hearing, or non-native speakers by providing live transcriptions and translations.

The project team itself comprises Microsoft employees fluent in over a dozen languages and dialects.

“This tool has the potential to significantly improve communication, and I am particularly excited about its ability to overcome language barriers,” stated Franklin Munoz, Principal Development Lead.

Data Privacy and Security

While the app is cloud-based, Microsoft emphasizes data privacy. It is not recommended for highly sensitive meetings. However, granular data and privacy controls are implemented, allowing users to manage their conversation data sharing preferences.

Data Handling and Improvement of Services

Audio and text data is transmitted to Microsoft’s online speech recognition and translation services, identified by a randomized identifier rather than personal information.

Meeting transcripts and recordings are saved locally on users’ devices. Participants are also given the option to “contribute” their meeting data to Microsoft to facilitate service improvements.

When all participants consent, Microsoft can retain audio and transcriptions for analysis. This data is used to refine speech recognition and speaker attribution accuracy. Access to this data is governed by NDAs and limited to Microsoft employees and contractors, excluding account credentials.

Reviewers only access randomized audio snippets, and Microsoft employs de-identification techniques to remove sensitive information like credit card or phone numbers. Users retain the ability to delete previously shared recordings, which are otherwise stored on encrypted servers for up to two years.

Considerations and Availability

Currently, there is no administrative control to disable the “contribution” setting for all users. Therefore, individuals should carefully evaluate the benefits and risks before using the service. It’s important to note that this is a Microsoft Garage project, implying an experimental nature and potential for discontinuation.

The Group Transcribe app is presently available exclusively on iOS platforms.

Early Stage is the leading event for startup entrepreneurs and investors, offering practical guidance on building and scaling businesses. Sessions cover fundraising, recruitment, sales, legal matters, PR, marketing, and brand development. Each session incorporates audience participation, allowing for questions and discussion.