Conversation transcription app or service?
January 16, 2023 10:27 AM   Subscribe

I have 12 hours of audio recorded from 8 focus group conversations. Can you please recommend me your favorite free or paid app or service for creating a transcription? Bonus points if it labels the different speaker names.
posted by rebent to Technology (10 answers total) 7 users marked this as a favorite
posted by chesty_a_arthur at 10:33 AM on January 16, 2023 [2 favorites]

I haven’t tried others but marks speaker changes and numbers them. It’s not perfect - if speaker 1 and speaker 3 sound similar, it might mislabel speaker 3. It’s relatively easy to fix and has been pretty good in my experience.
posted by kat518 at 10:35 AM on January 16, 2023 [1 favorite]

Best answer: I find Edit Eddy really good, currently free. It recognises Speaker A and B, I have not tried it with a group conversation though.
posted by Lluvia at 10:41 AM on January 16, 2023

Otter is generally great- it's had a few weird issues for me lately but mostly with recording, and since you already have the recordings it should be fine. I use it pretty much every day. It will label different speakers, although it sometimes doesn't distinguish well between people who have similar voices.
posted by pinochiette at 11:13 AM on January 16, 2023

I use regularly in my work, and I'm pretty sure you can do everything you need with the 14-day free trial. As mentioned above, if speakers sound similar it doesn't always label them correctly and overlapping dialogue can get garbled. These are usually minor issues that can be cleaned up manually and would happen with any service that doesn't use human transcribers. When there are many people in a conversation it becomes more likely you'll encounter these issues. If you need a very high degree of accuracy you can pay by the minute for a professional to transcribe your recordings.

With I think you'd have to pay for a month of the pro tier in order to upload pre-recorded audio (the free tier only does real-time transcriptions in meetings afaik).
posted by theory at 11:14 AM on January 16, 2023 [1 favorite] (which is not is more expensive but very high quality, with humans transcribing aided by AI. With the recent advances in automated methods, I don't know how much value it has (e.g. Whisper is free and as good or better than the models deployed by Amazon, Apple, MS, and Google with essentially human level performance).

One of the nice things about humans is that you can give them dis-ambiguating pointers that are shared context for the speakers (like common abbreviations or proper names). You can also tell them how many speakers to expect, and give time notes for speaker voices if you want to provide IDs
posted by a robot made out of meat at 11:42 AM on January 16, 2023

nth'ing - use it at work and, while not perfect, it's very good.
posted by NotMyselfRightNow at 12:45 PM on January 16, 2023

I am a fan of Sonix.
posted by Bella Donna at 1:50 PM on January 16, 2023

User #1 wrote a review of Descript that sounds like it does similar things to Edit Eddy.
posted by idb at 2:37 PM on January 16, 2023 [1 favorite]

Response by poster: I ended up using Edit Eddy. I liked how it was very simple - sign up, drag and drop audio, come back in a few hours and it's there. I also liked that it was FREE. If it becomes not free, I would be willing to pay for it on a per-file basis, it was pretty good for one-off support.
posted by rebent at 10:51 AM on January 23, 2023

« Older Ideas Databank   |   Advice on jumping a grade when moving school... Newer »
This thread is closed to new comments.