Quick Take Review: Otter.ai
Review of Otter.ai…Otter.ai transcribes audio conversations. The app performs real-time transcription with delayed corrections. It can also process existing audio files.
Basic version: Free with 600 minutes
Premium version $99.99 a year or $9.99 a month (6,000 minutes)
I have used Otter.ai in a number of situations, from noisy rooms at San Diego Comic-Con to telephone interviews conducted in a quiet home office. It works, but not perfectly. But when it comes to transcription, a good amount of text showing up without listening and typing greatly improves the interview and audio notetaking experience.
What Otter does
Otter listens to conversations and then transcribes them via automatic speech recognition (ASR). While Otter interfaces through iOS, Android and web apps, most of the processing takes place in the cloud. As the service transcribes the text, it also inserts punctuation and attempts to identify speakers. Otter records and processes conversations in near real-time, finalizing long sessions after a short post-input processing period.
The app also accepts audio files for transcription processing, including integration with the popular conferencing service Zoom.
What I like
Not listening and transcribing every word during an interview listen-back. It takes me forever to transcribe an interview. I listen back to the same sentence time and again. I don’t have the equipment for it, so I’m not very efficient. Outsourcing transcription isn’t cost-prohibitive in most cases but it does add time and cost to the process of publishing an interview.
The UI, on iPhone or iPad, allows the interviewer to edit the transcript in the app. I tend not to do this, preferring to export the text first and then edit in Microsoft Word while listening back to the audio.
I like that I can export to a variety of formats, including text, Microsoft Word and basic subtitles via SRT. The Premium version includes support for Dropbox.
When the Otter service works effectively, it allows the editor to identify the speakers. Once the service learns a speaker, it then assigns the name to that speaker’s comments throughout the transcript. With poor quality recordings this doesn’t always work.
The personal version also includes calendar integration with reminders for when to record meetings and capturing meeting information in the transcript header. Otter supports basic collaboration features like shared editing in the basic version. The Groups feature allows for the creation of a group of transcriptions, such as presentations from a conference. For those who want a visual peek into the transcribed text, Otter generates a word cloud on demand.
The premium version also includes custom vocabulary. This can be useful for business and science where the language of the work isn’t likely to be in the common set of words used to train the speech recognition engine.
While some features are restricted, I appreciate that Otter doesn’t include external ads in the free version, and I hope it stays that way. This app needs to be responsive and ads would get in the way of user experience and performance. Otter does include lightweight promotions for their own services in the UI.
What else does Otter do?
Otter also offers a service for Teams. The Teams version features shared speaker voice prints, collaborative editing, and conversations organized by folder. This type of service will be the future of collaborative sessions. If Otter performs as it does in the premium version, I doubt Otter currently capture all the dialog, as meetings can often become chaotic. Use of transcription software needs to resist changing the dynamics of meetings because of software limitations. If a meeting requires a roaring verbal feast with people talking over one another, don’t let transcription protocols get in the way of accomplishing goals.
The apps support pasting images into the flow to reinforce the transcribed conversation.
Be sure to test Otter in several “typical” meetings before committing to the platform as a solution, but do experiment and see if it fits your needs.
The Teams version also includes additional capabilities like user management, usage statistics, billing, and improved security.
Otter also offers an education version.
In our review of Otter.ai: room for improvement
In our review of Otter.ai, I found the service to make a lot of mistakes. That’s understandable on a poor recording, but even in a quiet space with a good recording, I still discovered missing sentences.
In one instance the service incorrectly assigned a speaker’s name from a previous conversation. I manually replaced the name and then set the service about rematching. Otter completed the task successfully save a couple of misses.
I would like to see the system create a pre-conversation profile, where speaker names can be input. If it knows me, it can assume the unknown speaker is the person in the profile. With more than one unknown speaker Otter could allow the person recording the conversation to select speaker assignment on the fly.
I would also like to see a sound profile metrics dashboard that lets me analyze the set-up ahead of a recording to see if the input is optimized for recording—and if not, at least set my expectation for an upcoming suboptimal session.
The pricing seems a bit high for this service, especially for casual users who want to use premium features. A good human transcription or two will, however, easily match the $99.99 annual fee. Casual users may well wonder about value in-between uses when they don’t even reach their monthly allotment of 600 minutes. For people who interview or transcribe audio notes regularly, the $9.99 a month fee, or $99.99 a year may be OK. I would like to see a lower tier that perhaps constrains the minutes but turns on features that aren’t compute-intensive, like exporting Word documents to the cloud. Also, if Dropbox support remains premium, the company should include the usual suspects like Google Drive, iCloud and Microsoft OneDrive as well.
Over time tools like Otter will get better. That is the nature of learning. The company, however, needs to address issues of picking voices out of noise and identifying areas of ambiguity. Perhaps assigning some probability as to correctness or completeness. Some level of semantic knowledge beyond pure machine learning could deliver higher quality outputs because it would be able to understand intent rather than just patterns. All of these suggestions will require investments in ancillary technology.
Review of Otter.ai: The Bottom-line
For those transcribing their own content, Otter.ai is much better than a manual process but not as good as a human transcription service.
For non-critical interviews where I have the time to make edits, Otter does its job good enough to remain a member of my technology portfolio.
Otter.ai is available on iOS, Android and on the web.
Read the recent Serious Insights report on mind mapping here.