In the evening, after a long day at work, I quickly type up the last of the day’s interview. As long as I still have the interview in my head, it should be quick – particularly given the next tasks are piling up on my desk or in my email inbox. I give up after two hours, and I haven’t even transcribed a third of the audio recording.
I know that somewhere help is out there in the form of tools that can convert spoken language into text. All within a few minutes, or at least in much less time than it takes me to transcribe text manually. Google’s certainly got the answer. I search for free tools and actually find what I'm looking for. So, I work my way through them but meet a few challenges.
As with many tools that can be used for free, free use is severely limited. Most of them allow me to transcribe up to 5,000 characters without taking out a subscription or charging me anything. I don’t get particularly far with that – a one-hour interview is considerably longer. With the next provider, I first have to download some software and I’m not quite sure whether even this process doesn’t contain hidden costs. I try it anyway, but here too the number of characters for a free transcription is so limited that I would have transcribed that same amount of text manually in the time it took to install the tool.
The quality of the free transcription tools varies greatly. Typically, it even begins to falter with the recognition rate. I end up with a spaghetti of words devoid of full stops and commas, let alone paragraphs, which needs reviewing against the audio and time-consuming corrections. The time I’ve saved is next to nothing.
Another disadvantage is that most tools simply turn spoken language into text without paying attention to small details – the use of numbers or dates, for example. The resultis ‘thirty-eight-point-two percent’ instead of ’38.2%’, which can be really tedious in the long run.
Aside from making work easier and transcription quality, there are other criteria that free tools tend to neglect – the issue of data protection, for example. In particular, interviews often involve confidential content – until the finished article has been shared by the interviewee. The extent to which sensitive data is protected from being accessed by third parties depends on the location of the servers on which the software runs and on the applicable data protection laws that regulate access rights to users’ data. German data protection law is among the strictest, but only applies if the provider’s servers are in Germany.
After testing various free tools, one thing is clear to me: if transcription software is really going to make work easier – that’s to say make work more efficient with high-quality results – it is worth investing in paid tools and paying attention to your provider.
For me, GoSpeech charges a fair price and can be used without a subscription. The speech recognition results are good. Post-processing means that certain terminology –time, length and medical terms, for example – is automatically corrected and written in the correct format. This saves me time. And, as this solution is exclusively hosted in Germany, it is GDPR-compliant. In other words, GoSpeech ticks all the boxes for the professional speech recognition of interviews.
You are a journalist and often work with audio recordings? Well take the opportunity to test GoSpeech for free now and have your audio and video recordings converted into text by our transcription software.