Whether you work in media, law, education, data analysis, finance, compliance, or simply need to record your Zoom meetings to read over later, you’re likely in need of a good transcription service. As a reporter, I rely on them on a nearly daily basis to turn hours of conversations with subjects and sources into accurate text files that I can use when writing articles.
Since there are at least a dozen services out there, ranging from companies that employ humans to ones that use AI, and even hybrids of the two, we decided to put 10 of the most popular services to the test.
For our test we used two two-minute iPhone recordings from a recent interview with Genius co-founder Mahbod Moghadam. One recording was clean, with no background noise, and the other was the same interview clip with added music with vocals in the background. We rated our favorite services based on the following criteria: pricing, ease of use, speed, features, and most importantly, transcription accuracy.
We came away with one clear winner, but all the services have pros and cons, like pricing and turnaround time, that could be worth your consideration. First, though, let's tell you about our winner:
Our top pick: Otter.ai
Otter’s other features, user interface, low price, and ease of use more than make up for a not-completely-perfect transcription, making it our top pick. Otter comes with team collaboration tools, not to mention a huge volume of free transcription minutes every month. Finally, the user interface is about as painless as it gets. Otter offers integration with Dropbox, Google, Zoom, and Microsoft, making it more versatile than other services. With the paid version, Otter Pro, users get a slew of extra features that sweeten the deal for only $13/month. Whether you’re a student or work in a newsroom, you’ll probably be able to afford Otter.
Our runner-up: Rev
Rev’s real strength is variety. With Rev, you can choose between a human or machine-generated transcription, while other services typically offer one or the other. Rev’s user interface is just about as intuitive as Otter’s. However, its higher price tag and less accurate machine-generated transcriptions keep it from getting the number one spot on our list.
If you want to dive into our research and rankings, here are our favorite transcription services, ranked from worst to best:
Pricing: $3.50/min for same day service; $1.25/min for 3-4 week turnaround; several pricing options depending on number of speakers and turnaround.
Transcription method: Human-generated.
Features: Timestamping, translation, rush orders.
Ease of use: GMR’s site is perhaps the most difficult to use. Just logging in leads to some password fill-in malfunctions that no other service seems to suffer from. After five failed attempts, users have to reset their passwords and try logging in again (which I’ve already had to do twice for this review). This is especially frustrating when trying to make payments. There’s also no transcript editor — all you get is a Word doc download.
Overall rating: While GMR’s transcriptions were completely accurate, its old-fashioned site makes the ordering process too difficult to be worth it. In fact, the transcription was a little too accurate — we don’t necessarily need the “right”s and “um”s in most transcriptions. Besides, other services, like Rev, add verbatim transcriptions as an option if that’s what you’re into. The lack of a transcription editor makes reviewing a file a lot harder, putting this service at the bottom of the list.
About the company: GMR’s customers include AT&T, Amazon, Chevron, Dell, and Microsoft. The company was founded in 2004, making it the oldest service on this list. GMR’s founder, Ajay Prasad, got the idea for the business after needing seven hours of audio transcribed and not finding a business with transparent pricing online.
Pricing: 6-12 hour turnaround: $2.50/min; 1-day turnaround: $1.30/min; 3-day turnaround: $1.10/min; 5-day turnaround: $0.90/min.
Transcription method: Human transcription.
Features: Captions, foreign subtitles, translation, rush order, timestamping, transcription editor.
Ease of use: While GoTranscript’s user interface is rather old-fashioned, it’s still got many of the same features as Rev, Otter, or other leading services. It’s not quite as intuitive, but by no means impossible to use.
Overall rating: Thanks to GoTranscript’s human-generated transcriptions, accuracy isn’t much of a problem. Both audio files were almost completely mistake-free, but the noisy audio sample had one or two mistakes. GoTranscript’s pricing changes depending on how fast you’ll need your transcription, making it potentially one of the more expensive options. Unlike Rev, GoTranscript doesn’t have an AI first draft ready and waiting while you’re waiting for a more complete version. If you’re not in a hurry and looking to save money on a transcription, GoTranscript’s 5-day turnaround option may be right for you.
About the company: GoTranscript was founded in 2006 in Edinburgh, Scotland, making it an early entry into the transcription market. To date, the company has transcribed 144 million minutes of audio, has had 14,000 customers, and according to its site, it boasts three office dogs.
Pricing: $0.89/min for 48-hour turnaround.
Transcription method: Human-generated transcription.
Features: Translation, subtitles, rush order, timestamping, verbatim transcriptions.
Ease of use: While iScribed’s website is easy enough to use, the lack of a transcription editor makes the service a little less user friendly. The site isn’t as archaic as some others, like GMR or GoTranscript, but it doesn’t have the same handy features as some of the more intuitive sites, such as Otter, Temi, Sonix, or Rev.
Overall rating: iScribed’s accuracy was nearly perfect, and the two audio files turned out identical transcriptions. However, there’s no transcription editor, which means major points off for user friendliness. This service would be ideal for those who want a cheap, fast, accurate transcript without the bells and whistles of other sites.
About the company: iScribed was founded in 2013. The company is headquartered in Lawrenceville, Georgia.
Pricing: $10 credit after signing up for newsletter, $0.10/min for machine-generated transcription; $0.80/min for human-generated transcription.
Transcription method: Machine-generated; human-generated.
Features: Transcription editor, timestamping, save to Dropbox/Google Drive, rush order.
Ease of use: Scribie’s transcription editor isn’t completely user-friendly — the font is awkward to read, and a separate window pops up for the editor. When listening to the audio, Scribie doesn’t offer captions either, making editing a slightly slower process.
Overall rating: Scribie’s human transcriptions are very accurate, while its machine-generated transcriptions had a couple of errors, especially when it came to the noisy audio file. However, at 10 cents a minute with a 30 minute turnaround, Scribie might be the best option for those looking for the absolute cheapest transcript.
About the company: Scribie was founded in 2008 by Rajiv Poddar, making it one of the first modern transcription services to enter the market. Poddar initially got the idea for his company while writing a recording plugin for Skype. Poddar started his own venture to address the lack of transcription services, which in his experience was a pain point for many customers.
Pricing: $0.79/min for first draft service; $1.25/min for standard service; $2/min for verbatim service.
Transcription method: Machine-generated transcription.
Features: Timestamping, transcription editor, rush orders, rough draft transcriptions, verbatim transcriptions, translation, AI datasets.
Ease of use: At the time of this review, TranscribeMe’s website was facing some technical issues on Safari. Google Chrome, however, loaded the site in seconds. While it’s not the fastest site in the world, its interface is user friendly, despite the lack of captions in the transcription editor.
Overall rating: TranscribeMe’s transcripts for both audio files were identical, and virtually mistake-free. This service arguably performed best with the noisy audio file. The slow site and price tag don’t necessarily help it stand out against similarly priced competitors like Rev.
About the company: TranscribeMe’s customers include Oracle, Kaplan, Facebook, Deloitte, and Salesforce.
Pricing: $50/month for standard plan (7 files per month); $75/month for advanced plan (unlimited); $85/month for pro plan (unlimited).
Transcription method: Machine-generated transcript.
Features: Transcription editor, chatbot, captions, timestamping, multi-user editing, enterprise plan for businesses, real time transcriptions, Zoom integration.
Ease of use: Trint’s user interface is extremely intuitive, and its transcription editor is very straightforward, formatted not unlike Google Docs.
Overall rating: Trint’s transcriptions weren’t quite as accurate as some other AI-based services, like Temi or Otter. The noisy audio file, oddly enough, had fewer mistakes than the clean audio transcript. While Trint has a user-friendly site, and a convenient monthly subscription as opposed to dollar-per-minute pricing, there’s not enough to set it apart from its more accurate machine-generated competitors, Otter and Temi.
About the company: Unlike other companies on this list, Trint was founded by a former reporter, Jeff Kofman, who had been frustrated by the lack of good transcription services for years. In 2014, Kofman founded the company, and to date, it’s raised over $14 million from the likes of the New York Times and Google’s Digital Innovation Fund.
Pricing: 30 mins free transcription; $10/hour for standard service; $22/month plus $5/hour for premium service.
Transcription method: Machine-generated.
Features: Transcription editor, translation, subtitles, collaborative tools, integration with Zoom, Google, Cisco, and Adobe, timestamping, captions, editing tutorials, chatbot.
Ease of use: Sonix’s site is extremely user friendly, and so is its transcription editor. Unlike some other services, Sonix has captions in its transcription editor, which makes fixing errors much easier.
Overall rating: Sonix’s transcriptions are relatively mistake-free, but not perfect. The clean audio sample had slightly fewer errors when it came to inaudible bits of dialogue and speaker attribution. However, the noisy audio transcription featured a pop-up message to explain that some editing was required, and which areas of the text were less accurate. Sonix’s user interface, editor, and affordability place it relatively high on this list.
About the company: Sonix’s customers include NBC, ABC News, The New Yorker, Vice, and Google. Jamie Sutherland, Sonix’s co-founder and CEO, is a serial entrepreneur and software executive. Sonix’s other co-founder, David Dat Nguyen, was previously one of Gusto’s first employees.
Pricing: One free transcription for new accounts; $0.25/min.
Transcription method: Machine-generated transcription.
Features: Time stamps, captions, transcription editor.
Ease of use: Temi’s turnaround time is extremely fast, and its transcription editor is just about identical to Rev’s, making it extremely easy to use.
Overall rating: Temi’s transcriptions aren’t as accurate as Rev’s human transcriptions, but they’re better than Rev’s machine-generated transcriptions. Our clean audio sample had almost no errors, while the noisy audio sample had a handful, meaning you might have to do some light editing once you receive your transcription. If you’re looking for a cheap, speedy, relatively accurate service, Temi is your best bet.
About the company: Temi is actually affiliated with Rev, which would explain why their user interface and transcription editor is identical. While Rev’s focus is more on human-generated transcriptions, Temi’s is AI only. Temi’s over 10,000 customers include businesses like ESPN and the San Diego Union-Tribune.
Pricing: Transcription: $1.25/min; $0.25/min for rough draft.
Transcription method: Human transcription for standard service; machine-generated for rough draft.
Features: Machine-generated rough draft, time stamps, captions, foreign subtitles, live captions for Zoom, transcription editor.
Ease of use: Rev’s interface is extremely intuitive and easy to use, and it offers playback that syncs with text.
Overall rating: We found that Rev’s human-made transcript was by far the most accurate, both with our clean audio and background noise audio samples. While the machine-generated transcript arrived within minutes, it couldn’t get around the crosstalk and less audible sounds from the noisy audio sample. However, it’s ideal for those who only need a quick, rough transcript at a fraction of the price of a regular one.
About the company: Rev was founded in 2010 by MIT grads Dan Kokotov, David Abrameto, Jason Chicola, Josh Breinlinger, Mark Chen, and Paul Huck. To date, the company has raised $33 million in funding and has 170,000 customers. Rev employees around 50,000 freelancers for its human transcription, and has a full time staff of 150. In addition to being one of the most popular transcription services, Rev is PCMag’s top pick, and Forbes listed Rev as one of its top 50 AI companies of 2019.
Pricing: 3 free imports; 600 mins/month for free plan; 6,000 mins/month for Otter Pro ($13/month).
Transcription method: Machine-generated transcription.
Features: Free plan: timestamping, transcription editor, captions, calendar sync, Zoom and Google Meet sync; Otter Pro: custom vocabulary to improve transcription, video and audio sync with Dropbox.
Ease of use: Otter’s user interface is extremely intuitive and easy to use, and includes several useful widgets for sharing transcriptions with others, whether it’s via email, Google or Microsoft calendars, or Dropbox. Otter’s interface also includes groups and folders for transcriptions, making its site not only a place to transcribe, but a convenient hub for all your transcriptions. For those who aren’t familiar with transcription services, Otter offers tutorials for their site.
Overall rating: Otter’s transcriptions were accurate for the most part. The clean audio sample was almost mistake-free, while the noisy audio had a handful more, but was still coherent. Using Otter’s vocabulary feature may help improve the AI the more you use it. Otter’s a great choice for those who have a large volume of transcriptions to get through, or for teams looking to collaborate remotely.
About the company: To date, Otter has transcribed over 100 million recordings, adding up to over 300 billion total minutes. In 2020, Otter’s revenue was up to eight times what it was the year before, likely due to COVID-19. The company has $23 million in funding, and was founded in 2016 by Sam Liang and Yun Fu. Investors include Slow Ventures and Tiny Capital.
