A journalist, on average, spends 6 hours a week transcribing audio, which not only slows down the reporting process but also decreases newsroom productivity. Over the course of a year this equates to 2 lost business months which, at a time of profound cutbacks across newsrooms worldwide, makes the case for automated transcription services compelling.
With advances in AI, voice recognition and algorithm tech, the standard and quality of automated transcription services has reached a tipping point. So why are publishers still painfully transcribing interviews by hand, costing both time and resource?
It turns out, publishers don’t need to. Automated transcription tools are finally getting good … as in really good.
The best services have now evolved away from human transcription and rely instead on a simple premise: simply upload an audio or video to a site (or software) and within minutes, receive a fully automated, comprehensive transcript.
The current challenge is the sheer volume of transcription services now becoming available, with new AI-powered entrants making quick, rapid and disruptive moves into the market. This is tempered by the fact that many are start-ups who are either under-resourced or have glitchy tech.
WNIP has scouted the transcription landscape to come up with what we believe are the top five transcription services available in the world today. We based our findings on anecdotal evidence, reviews, and crucially the quality and ease of use of the service in front of us.
Here are our top five:
Dragon speech recognition software is available on both Windows and Mac (and mobile) and claims to be 3x faster than typing and 99% accurate. It lives up to its stats.
Straightforward to set up, quick to configure and easy to learn (it’s ready straight out of the box), users can expect significant productivity gains with automated transcription, whether it’s from pre-recorded audio files or podcasts.
Developed by Nuance Communications out of Massachusetts, the software has three primary areas of functionality: voice recognition in dictation with speech transcribed as written text, recognition of spoken commands, and text-to-speech: speaking text content of a document. The Pro version allows creation of custom commands to control programs or functions.
Its chief drawback is that it won’t work for multiple speakers or to transcribe presentations and lectures. It’s not cheap either, weighing in at a hefty $300 for the Pro Version, Mac v6.
Need something that’s not platform dependent? Welcome to Trint. The web app takes your audio/video and automatically generates a transcript in just a few minutes. Transcription quality is acceptable and generally requires only minor punctuation editing. It’s well-designed, intuitive, and allows you to export transcripts in various formats.
It’s chief USP is that it can listen to an audio recording or a video of two or even multiple speakers and provide a written transcript of what was said. Unlike Siri or Google Talk, Trint is designed to transcribe long blocks of text. It also transcribes in all the major English accents, plus 12 European languages.
Launched in London during 2016, Trint’s costs depend on usage, starting at €16.20 per hour for Pay As You Go, rising to €130 for ten hours of upload a month. Whilst it is still a nascent technology, it’s already a favourite of WIRED and was part-funded by Google Digital News Initiative.
Used by over 4,000 journalists and researchers (including Poynter, USA Today and Forbes), Happy Scribe, is an established automated transcription service that works across more than 119 languages and accents. It transcribes audio files in just a few minutes, usually taking around half of the file length in time.
The company is still relatively new and formed part of the DCU UStart programme and The Ireland Funds Business Plan. Co-founded by two Spanish and French graduates, it was chosen by Silicon Republic last year as one of their startups of the week with a nod towards its ease of use, pricing model and the quality of its transcriptions.
Whilst Happy Scribe’s Interactive Editor is a strong example of smart intuitive tech, its pricing model starts at just €6/hour on a pay as you go basis. As such it’s become a fast favourite of publishers with rapid uptake since its launch.
Cassette is an Indian-developed iPhone app which allows publishers to seamlessly record, transcribe and search important conversations on the fly. It’s as much a voice recorder as a dedicated subscription service and, with 90% transcription accuracy, is a superb resource for journalists on the move.
One of Cassette’s best features is that it helps find what’s important, searching across all your recorded content to “find any quote, note, conversation, or project you’re looking for instantly.” Briefs is another unique feature — automatically generated short visual summaries of your conversation’s highlights. Essentially a concise overview of what matters.
Whilst the app is free, the transcription services are not. Monthly plans start at $10 for 3 hours transcription going all the way to $125 for 50 hours transcription.
Recordly takes Cassette’s features one step further by even integrating smartwatches into the process of recording and transcribing audio. Whether that’s a key USP is a moot point, but it’s indicative of Recordly’s forward-thinking app that acts as a ‘reporting ecosystem’ with the smartwatch acting as a remote control to highlight important parts of recordings.
The app relies on the power of IBM’s Watson AI platform to generate its transcripts with the app sending push notifications to users once the transcript is complete. The transcription services themselves start at $2 per hour of audio.
Its chief drawback is that it’s iOS-only, at least for now, and whilst an Android web app is in the pipeline, the service is currently limited to Apple users.
In conclusion, whether you are using a Windows desktop or an Apple Watch, there’s a cutting-edge transcription tool to help you. Costs are falling and entry points are affordable for nearly every publisher. Moreover, with the rapid advancement of AI technology, we can expect to see some revolutionary developments in transcription over the coming months and years.
Our current favourite? It would have to be Trint, which has been described by Google as ‘rocket-fueled innovation’ and by WIRED as ‘amazing’. In saying that, all the above transcription services are cutting edge and have different features and benefits publishers can benefit from. It’s simply a case of evaluating which offering is best suited for your newsroom.