Speech recognition in real time
Real-time speech recognition technology has revolutionized the way we consume audio and video content. By transcribing spoken language into text, it is now possible to provide real-time captions for live broadcasts, making them accessible to a wider audience, including those who are deaf or hard of hearing. ASR (Automatic Speech Recognition) technology has come a long way in recent years, and is now capable of accurately transcribing spoken language in real time with high accuracy. This has enabled broadcasters to provide captions for live events, news broadcasts, and other programming, making it easier for everyone to follow along and stay informed. Moreover, captioning can also benefit people who are learning a new language or who are watching videos in noisy environments where it may be difficult to hear the audio clearly.
How does it work?
The event broadcast director can utilize the Real-Time Messaging Protocol (RTMP) to transmit the event stream to the Tiro real-time captioning server. After processing, the Tiro server returns the stream with the events captioned text in Icelandic. This stream can then be forwarded to any streaming service that supports RTMP, providing accessible captions to the audience.
Showcase
More to come:
This solution is new, and we are excited to continue improving it for our users.
- The stream's text is automatically sent to the captioning system on talgreinir.is, where SRT files can be downloaded or the text can be further edited.
- The text can be displayed in embedded captioning systems in web players, but this depends on the implementation of each individual web player.
- Broadcasters can switch to other languages if the speaker is speaking in a language other than Icelandic, for example, English.
- It is possible to increase the weight of certain vocabulary, which can ensure that domain-specific names and vocabulary that may be outside the conventional language are correctly transcribed.