A free, easy way to convert speech to text

Speech recognition has come a long way. I should know. Back in 2001 or so I bought speech recognition software and a headset, hoping to do away with typing all together and simply dictate my writing. That didn’t work out well because the technology was in its infancy. The transcriptions were garbled and barely usable – and it took far longer to clean them up than it would to have typed it in the first place.

So, I took the sensible step (one I should have done decades earlier) and taught myself to touch type.

However, there are still plenty of times when I wish there was a quick and easy way to transcribe conversations into text. I suspect many writers feel the same way: they may produce podcasts and want to provide a transcription; or may interview people to gain opinions and information for a book and don’t want to spend hours typing it up.

It’s not only professional writers who can benefit of course. Anyone who uses the internet much these days may come into situations where they’d benefit from a text version of some spoken audio.

If you need frequently need professional standard transcriptions, then it probably makes sense to pay up for a dedicated service or software. However, for more modest requirements, the free and easy way to transcribe audio is to convert it to a .wav file, and then run it through IBM’s free to use Watson service.

Speech to text transcriptions – the free option

Watson is a question answering computer system. It uses IBM’s speech recognition capabilities to convert speech into text, with the  transcription provided almost instantly. It works with a range of languages too, including: US English, UK English, Japanese, Spanish, Brazilian Portuguese, Modern Standard Arabic, and Mandarin.

It can work with live audio, or with recorded files. So next time you want a conversation transcribed, give Watson a try. The service is in demo at the moment (February 2016) – so availability could change in the future. But at present, it’s free to use for 1,000 hours a month. Since even the longest of months contain only 744 hours that should be enough for most people!

If you try the service, let me know what you think in the comments.