Well we can’t answer that for you (exactly) but we can review and compare them using Episode 166 as the common sound file. The goal was to answer (or try too anyways) the often asked question – what service should I use to transcribe my podcasts?

There are however many caveats which we cover during the episode, such as:

  • How many speakers do you have?
  • Do you have accents?
  • Do you need a vocabulary built in?
  • What is your budget?
  • How long are your episodes?
  • How often do you record/publish?
  • What is your required turn around time?
  • Do you want to do any editing?
  • Do you care if it is perfect?
  • And many more!

So just keep in mind that the tests below are based on a single 17 minute file with 2 speakers. Your mileage may vary with each service depending on how you answer any of the above questions. Or what I am trying to say is – use this episode and our transcription service research to help you validate or shorten your final list.

Transcription by Human vs. Transcription by AI

So which is better? Human transcription is likely going to be more accurate today but also much more expensive. Based on the services we looked at AI can run from $.40 to $5 or so to transcribe a 20 minute audio file. The services that have humans transcribe ranged from $20 to $30 for the same file. While we didn’t compare AI vs. Human for this test from what Dave has seen, human transcriptions are much closer to 99.99% correct mark than the AI options are – for now.

So in the end your options are:

  • Be cheap and use AI.
  • Go for accuracy and use a human service.
  • Be cheap and use AI but spend time editing it after the fact.

With that said we mostly looked at AI services but did include some often mentioned and popular human transcription service options too in the review section.

Transcription Service Scores

ServiceCost Per 20 Min File
Average Score
Ease of UseFeaturesSpeaker HandlingTranscription
Descript (free)$0.007.757798
Descript$0.407.757798
Otter AI (free)$0.0078587
Otter AI$0.0378587
MS Word$0.006.507397
Temi$5.006.507298
Audext$3.635.251497
Sonix AI$3.3356527
Amazon (free)$0.004.752296
Amazon$4.804.752296
Happy Scribe$5.154.756526
Simon Says AI$5.004.256650
Trint$6.864.253626
InScribe$1.00112N/A0
Cassie Hauschildt$20.00N/AN/AN/AN/AN/A
Rev$25.00N/AN/AN/AN/AN/A
Go Transcript$26.00N/AN/AN/AN/AN/A
GMR Transcription$30.00N/AN/AN/AN/AN/A

Regarding the scores, here is some quick background on each column.

  • Ease of Use – Is it easy to setup and use? Is it fast? Does it work?
  • Features – Does it have integrations (Zoom, Zapier, etc.)? Does it have export options? Can I get in there and edit if needed?
  • Speaker Handling – Does the service label speakers? Seems simple but many failed.
  • Transcription – I looked at 4 different clips and compared to reality and other services. If major issues were noticed they lost .5 or 1 point per.

Transcription Service Reviews

The reviews were done by Dave over the period of 1-2 weeks in August/September of 2020. All pricing, features and such are based on what was available on the sites, apps and such at that time.

Descript (Score: 7.75)

  • Free Option – 3 hours of transcription (one time)
  • $12 a month gets you 10 hours of monthly transcription

Descript is a bit different as it is an APP that you have to download. Also to be fair it is what we currently use for our transcripts so note that I am more familiar with the program than a newb. Once you download and get setup the Descript App is pretty easy to use. I simply create a new Composition in my usual folder, move the file in and let it go to work. I set assign speakers and once it is done it lets you assign the various voices to a person and then updates everything. The result is usually 100% match on speakers, even if 3 or 4 people it seems to work. Only once in some 100 or so episodes have I had issues with the speaker assignment. The output is clean and simple and overall I do love the app and price. That said it will be interesting to compare the quality of the output to others finally.

Otter AI (Score: 7.00)

  • Free Option – 600 minutes a month (max 40 minute file)
  • $8.33 a month gets you 6000 minutes a month (max 4 hour file)

Otter has a simple signup with the usual email verification. Out of the gate you can upload 3 files and get 600 minutes even if you don’t sign up for a plan. At the free level the integrations are limited but if you upgrade Zoom and Dropbox become available – no Zapier however. There is also a Manage Vocabulary – only 5 words with free but again if you upgrade you can have 200. So for those with technical terms, industry terms or difficult last names this again can help improve the accuracy. One interesting feature that others don’t seem to have is contacts and calendar integration. Included below is a screen shot of the app pushing to help you tie into your meetings and basically be a note taker for you. Lastly the export is somewhat limited compared to other tools. If you are free you are very limited but even if paying the options are lagging other tools. The processing also took much longer than any other solution when you factor in that it has to process, then while working I assigned speakers and then it finished the transcription but still took awhile to assign speakers. So for a 16 or so minute file and compared to other AI tools it was the slowest.

Microsoft Word (Score: 6.50)

  • Price: Free with online 365 office suite

I think the hardest part of using this solution was realizing that it isn’t in Word on my laptop but rather only available in the browser version of Word for Office 365. So if you have Edge browser or Chrome and can log into your 365 Office account you can use this tool. As far as the use and output, both are really straight forward and simple. I just selected a file, waited and then boom its done. You can then edit as needed in a Word doc in the browser or locally after you export. A bonus was that while you cant assign speakers it does list Speaker 1 and Speaker 2 so after a simple Search and Replace you can have speakers assigned – not something most other systems do or allow oddly.

Temi (Score: 6.50)

  • $0.25 a minute

So Temi is by far one of the easier solutions to get started with we used. Right on the homepage they let you upload a file and then ask for your email and go right to work. For a lead generator but also making it fast and easy to test their service they might be the best. The interface once you get your file is really simple. You can do some edits, change speakers (I had to change one early on that the AI got wrong) and can have it edit out uhs and ums. Export has some options to do the same and all in all is really fast and simple.

Audext (Score: 5.25)

  • Hourly – $10.90-13.00 per hour. Plans $8.99-12.00 with more savings if go annual.

I was able to sign up but the email – never came. I checked JUNK as it kept suggesting but never saw it. I then tried other email addresses in an attempt to just start over. Too late, my IP address was now attached to the account I couldn’t get into or recover. I cleared cookies and changed my IP and was able to finally get in using a GMAIL account. Then my troubles continued. I tried to upload a file and it simply took forever. Eventually it timed out and I hit back and tried again. Timed out again. I did this 5 times and I guess it DID work every time but I had no idea. So I blew through my 30 minutes and had 3 more of the same file waiting to be transcribed if I paid. So it worked then? It did but I had no idea. Audext also has a unique feature where ahead of time it asks for the speakers to be listed and then uses those names in the output. As most failed to properly assign speakers or didn’t assign them at all I am impressed that the tool appears to do it so well. The final view is clean and simple but exporting is pretty basic. You can either go with txt file or .doc file and that is it.

Sonix AI (Score: 5.00)

  • $10 an hour or $22 + $5 an hour

Sonix is interesting but does have some great integrations setup (Zoom & Zapier), Google Chrome Extension and API access too. The handling of Speakers though was pretty poor in my test. From what I can tell it didn’t label Speaker 1 and Speaker 2 and I was manually going to have to do that. In the last name test it also was pretty far off for both of us. For me it said Raw… not even roar. Matt’s was interesting as well. The backend is pretty elaborate though and lets you highlight words it isn’t sure about, do translations and some other features that personally i wouldn’t use (but maybe you would).

Amazon Transcribe (Score: 4.75)

  • Free – 60 minutes a month for 12 months
  • Paid – is $0.0004 per second i.e. $0.24 a minute

Quite the setup Amazon Transcribe is. I already have AWS setup so those steps were already done for me. Then I simply had to turn on Transcription and upload something to S3. Sadly the files have to be locally hosted on AWS somewhere in a bucket for the system to work. There is an API so once setup you could avoid some steps. Once everything is setup (the most complex series of steps of any the solutions) you need to take your time running the transcription. Luckily you get 60 minutes a month because I used 17 and forgot to turn on speaker identification. This made it so I had to run another transcription. The export is in JSON so if you are hoping for something that is easy to get in Word, txt, or some specialty file – not this time. Honestly – after 3 tries I still had settings wrong and couldn’t get a full clean export but had managed to turn on the removal of any PI data (personal identifiable). Fast but man it isn’t easy to get things setup and going.

Happy Scribe (Score: 4.75)

  • 1 hour = 12 pounds (as in UK currency)

HappyScribe only lets you test the first 10 minutes and even though the test file was 17 it just simply stopped at 10 minutes. Handy! The control area is simple and clean. The process was fast and the turn around was pretty quick for the 10 minutes of transcribing. You can create your own vocab and use it – as with all we didn’t do that but there is that option. Downside – multiple speakers. This is another tool that makes you listen and set the speakers so if that matters to you know that you will have some work to do here. Lastly the export is simple and you have a ton of options to pick from in formatting and speakers/timestamps/etc. Do note that pricing is in pounds and not USD or Euros even.

Simon Says AI (Score: 4.25)

  • $15 an hour if pay as you go, Starter is $15 a month plus $7.5 an hour

Well Simon Says seems to only give you 15 minutes to try and they mean it. The test file was 17 minutes and to run the test I was going to have to run a credit card for the $2. While I shop online I really wasn’t interested in the paperwork and hassle for the test to take the $2 charge. Thus, there is no comparison on the quality compared to others. The cost overall compared to other AI options per minute does seem to be high. The admin area has some quick forced tutorials that are just a few clicks but helpful.

Trint (Score: 4.25)

  • 84 files a year @ $48 a month. Unlimited @ $60 but these prices are for annual pricing.

The Vocab Builder is an interesting feature that some others don’t seem to have. If you have some technical terms (or last names) that often require editing post transcribing, this feature might really help you. Trint also offers integrations with Zoom and Zapier which seems to be the least that many services have (if they have any). One thing I noticed while testing was a version history. With Sonix I made a mistake when dealing with speaker assignment and didn’t see a clear way to undo so for those editing in the app again this might be helpful. My biggest complaint with Trint is the handling of multiple speakers – it puts the onus on me to assign who is who – a big time sink if you ask me. The trial is weird and I may update this review. You have 7 days to cancel and after I did on day 1 the display was a bit scary in how it made me think I had already been billed.

InScribe (Score: 1.00)

  • 14 day trial and $99 to buy software after that.

Well I had high hopes for InScribe and installed it without adding QuickTime. I couldn’t get it to work so I closed it and installed QuickTime. The application still wouldn’t transcribe for me on my Windows 10 PC. Maybe it works better on a Mac? There was a great guide that answers a ton of questions and gives lots of help but nothing like a video or much about Windows 10. From what I can tell I do think multiple speakers would be an issue for this solution as well so for me personally this and all solutions that fail to make it easy to tag speakers gets a downvote. (Note that pricing was based on $99 charge and assuming that you do 40-50 episodes a year for 2 years)

Human Powered Transcription Services

To be fair these were not put through the same testing as the AI services so these are not “complete” reviews by any means. However we did want to highlight and mention at least a few for comparison to the AI options. The price per hour also is really variable depending on number of speakers, length, rush or no rush job, etc.

Cassie Hauschildt Transcription Services

  • $1.00 a minute

Pricing can go down for various reasons – size of project, size of company, etc.

Rev

  • $1.25 a minute

Depending on really anything the price just keeps going up. Want it faster? Up. Want it really fast? Up. More speakers? Yup, up in price.

Go Transcript

  • $1.30 a minute

Price can go up for rush but also goes down if you are open to waiting longer and buy more minutes. They do offer a one file/time free trial though.

GMR Transcription

  • $1.50 a minute

Price can go up for rush, more than 2 speakers and if audio is poor along with other reasons.

Transcription Sample Comparisons

Happy Scribe and Amazon Transcribe were hit with penalties since they did not seem to transcribe past the 10ish minute mark. Happy Scribe has a 10 minute limit on their trial so this may or may not be fair but it is our review! Amazon Transcribe however we aren’t sure why or where it stopped. We used the same sound file for all but yet it didnt export or transcribe it all.

Beyond that the one thing we learned from comparing the 4 samples from each service is that you really should create your own dictionary with the service. If you use names or acronyms they all should be in that dictionary – if nothing else is.

Intro Transcription Sample Comparison

So here are the services that we were able to test the introduction for and the results. Basically they are all the same except for punctuation and last names.

  • Descript – Welcome to another exciting episode of the business of digital podcast, featuring your host, Matt and Dave roar.
  • Otter.ai – Welcome to another exciting episode of the business of digital podcast, featuring your host, Matt Silva and Dave Rohrer.
  • Microsoft Word – Welcome to another exciting episode of the business of digital podcast featuring your host Matt Siltala and Dave roar.
  • Temi – Welcome to another exciting episode of the business of digital podcast, featuring your host, Matt silt and Dave roar.
  • Audext – welcome to another exciting episode of the business of digital podcast featuring your host, Matt Silsila and Dave Roar.
  • Sonix AI – Welcome to another exciting episode of the Business of Digital podcast featuring your host, Matt Ciotola and Dave Raw.
  • Amazon Transcribe – welcome to another exciting episode of the business of digital podcast featuring your host, Matt Silsila and Dave Roar.
  • Happy Scribe – Welcome to another exciting episode of the Business of Digital podcast featuring your host, Matt Ciotola and Dave Raw.
  • Trint – Welcome to another exciting episode of the Business of Digital podcast featuring your host, Matt Ciotola and Dave Raw.

Marker 1 Transcription Sample Comparison

A few services heard SEO as Damasio, Nescio, or “a messy oh” but some did get SEO. Audext heard API as “a p, I” which seems really odd but perhaps it doesnt understand the acronym? The one thing that jumped out to me is that you really need to create a dictionary of acronyms and names for your show – no matter what service you use.

  • Descript – Dave Rohrer: [00:02:00]…..And it wasn’t that really easy, you know? And everyone’s like, don’t do this. This is dumb. Use an API, change your CMS. Um, you know, you’re doing it the wrong way. Creating this manually is horrible. But no one actually asked, what is the business reason for this? Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone, you know, is there actually a business? And I think so often, um, Um, and it’s not just Damasio people and PPC, but so often people are like, when they look at why websites do things, if you want….
  • Otter.ai – Dave Rohrer 1:00 ….. And it wasn’t that really easy, you know? And everyone’s like, Don’t do this. This is dumb, use an API, change your CMS. You know, you’re doing it the wrong way, creating this manually is horrible. But no one actually asked, What is the business reason for this? Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone, you know, is there actually a business and I think so often. It’s not just an SEO people and PPC, but so often people are like, when they look at why websites do things, if you want….
  • Microsoft Word – Speaker 2 And we would always get something like Oh, we had it. We tested it still not working. OK, well next Sprint right? Yeah sure. Three months later we finally get it like Sprint through every three weeks or so. Come on. Yeah, it’s like. Speaker 2 Ask the questions but. Speaker 2 And there’s just so much going on for companies and internally resources and stuff. I think it’s hard to just judge. Speaker 2 Maybe that’s about quit judging peoples online.
  • Temi – Speaker 2 (02:12): And it wasn’t that really easy, you know? And everyone’s like, don’t do this. This is dumb. Use an API, change your CMS. Um, you know, you’re doing it the wrong way, creating this manually is horrible, but no one actually asked, what is the business reason for this? Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone, you know, is there actually a business? And I think so often, um, um, and it’s not just Damasio people and PPC, but so often people are like, when they look at why websites do things, if you want….
  • Audext – 00:02:10 – 00:02:55….and it wasn’t that really easy. Oh, and everyone’s like, Don’t do this. This is dumb. Using a p, I change your CMS. Um, you know, you’re doing it the wrong way. Creating this manually is horrible. But no one actually asked What is the business reason for this? Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone you know. Is there actually a business? And I think so often. Um um it’s not just a messy Oh, people in PPC But so often people are like when they look at why websites do things
  • Sonix AI – [00:02:04] …. And it wasn’t that really easy, you know, and everyone’s like, don’t do this. This is dumb. Use an API, change a mass. You know, you’re doing it the wrong way, creating this manually is horrible, but no one actually asked, what is the business reason for this? [00:02:32] Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone you know, is there actually a business? And I think so often it’s not just Nescio people and PPC, but so often people are like when they look at why websites do things. If you want ….
  • Amazon Transcribe – Speaker 0:….and it wasn’t that really easy. Oh, and everyone’s like, Don’t do this. This is dumb. Using a p, I change your CMS. Speaker 0: Um, you know, you’re doing it the wrong way. Creating this manually is horrible. But no one actually asked What is the business reason for this? Speaker 0: Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this Speaker 0: by management or by someone you know. Is there actually a business? And I think so often. Speaker 0: Um um it’s not just a messy Oh, people in PPC But so often people are like Speaker 0: when they look at why websites do things Speaker 0: if you want. ….
  • Happy Scribe – Dave Rohrer: [00:02:00]….And it wasn’t that really easy, you know? And everyone’s like, don’t do this. This is dumb. Use an API, change your CMS. Um, you know, you’re doing it the wrong way. Creating this manually is horrible. But no one actually asked, what is the business reason for this? Like, why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone, you know, is there actually a business? And I think so often, um, Um, and it’s not just Damasio people and PPC, but so often people are like, when they look at why websites do things, if you want, …..
  • Trint – [00:02:02] …..And it wasn’t that really easy. And everyone’s like, don’t do this. This is dumb. Use an API change or see a mass. You know, you’re doing it the wrong way. Creating this manually is horrible. [00:02:27] But no one actually asked, what is the business reason for this? Why are you doing it this way? What is the upside? You know, you’re being told that you have to do this by management or by someone you know. Is there actually a business? And I think so often. And it’s not just Nescio people in PPC, but so often people are like when they look at why Web sites do things if you want. ……

Marker 2 Transcription Sample Comparison

I think my favorite thing here is how Trint heard CMS as EMS and CMS as mess. Close but not really the same. Oddly no other service did that. Audext made up a word for “blog” by transcribing “blawg” which is an odd one. Otherwise really all the services were very similar.

  • Descript – Dave Rohrer: [00:08:00] ….I had to fight and show business cases and, you know, Examples of how all of our competitors are doing this and how we were getting destroyed in links and content, because we couldn’t do that stuff. We couldn’t move quick enough. And I had to slowly build a case to get a CMS, to [00:09:00] get a blog, to get certain things. So, you know, this one example was he was trying to build, you know, a site map. Someone’s like, we’ll get a better CMS that…
  • Otter.ai – Dave Rohrer 7:47 …..I had to fight and show business cases and, you know, examples of how all of our competitors were doing this and how we were getting destroyed and links and content because we couldn’t do that stuff. We couldn’t move quick enough. And I had to slowly build a case to get a CMS to Get a blog to get certain things. So, you know, this one example was he was trying to build, you know, a sitemap someone’s like we’ll get a better CMS that….
  • Microsoft Word – Speaker 2 …..I had to fight and show business cases an you know. Speaker 2 Examples of how all of our competitors were doing this and how we were getting destroid in links and content because we couldn’t do that stuff. We couldn’t move quick enough and I had to slowly build a case. Speaker 2 To get a CMS to get a blog to get certain things. Speaker 2 So you know this one example was he was trying to build, you know, a site map. Someone like will get a better CMS that.
  • Temi – Speaker 2 (08:41):….I had to fight and show business cases and, you know, examples of how all of our competitors are doing this and how we were getting destroyed in links and content, because we couldn’t do that stuff. We couldn’t move quick. And I had to slowly build a case to get a CMS, to get a blog, to get certain things. So, you know, this one example was he was trying to build, you know, a site map. Someone’s like, we’ll get a better CMS that……
  • Audext – 00:08:20 – 00:09:01 …..I had to fight and show business cases and, you know, examples of how all of our competitors were doing this and how we’re getting destroyed in links and content because we couldn’t do that stuff. We couldn’t move quick enough, and I had to slowly build a case to get a CMS to get a blawg to get certain things. 00:09:02 – 00:09:40 So, you know, this one example was he was trying to build, you know, a site map someone like would get a better CMS that …..
  • Sonix AI – [00:08:42] I had to fight and show business cases and, you know, examples of how all of our competitors were doing this and how we were getting destroyed and links and content because we couldn’t do that stuff. We couldn’t move quick enough. And I had to slowly build a case to get us CMS to get a blog to get certain things. So, you know, this one example was he was trying to build, you know, a site map. Someone’s like, we can get a better CMS that.
  • Amazon Transcribe – N/A
  • Happy Scribe – [00:08:42.680] I had to fight and show business cases and, you know, examples of how all of our competitors were doing this and how we were getting destroyed and links and content because we couldn’t do that stuff. We couldn’t move quick enough. [00:08:55.790] And I had to slowly build a case to get us CMS to get a blog to get certain things. So, you know, this one example was he was trying to build, you know, a site map. [00:09:07.760] Someone’s like, we can get a better CMS that. …
  • Trint – [00:08:42] I had to fight in show business cases and, you know, examples of how all of our competitors were doing this and how we were getting destroyed and links and content because we couldn’t do that stuff. We couldn’t move quick enough. And I had to slowly build a case to get us E.M.S. to get a blog, to get certain things. So, you know, this one example was he was trying to build, you know, a site map. Some like we’ll get a better see a mess that….

Outro Transcription Sample Comparison

Not sure why but Amazon Transcribe didnt fully work. Happy Scribe cut out the free trial at 10 minutes so we cant compare them. Otherwise really words like “Alrighty”, first and last names, and company names are really the only differences in terms of transcriptions. The things to note however is the time stamps and speaker setting (or lack there of) as if you want more detailed transcription and one that properly assigns a speaker. Some do a great job and others fail at assigning these well. Who is speaking here Trint & Sonix? What time is this Otter and Microsoft?

  • Descript – Matt Siltala: [00:16:37] Alrighty. Well, um, make sure that, uh, you’ve given us a five star review on iTunes. If you have a chance opportunity. Go look at our page. You’ll see all the places where we’re listed, uh, for Dave war with Northside metrics. I met salsa with avalanche media. Thanks guys. Bye. I met salsa with avalanche media. Thanks guys. Bye.
  • Otter.ai – Matt Siltala 15:43…….. Alrighty, well, you guys. Make sure that you’ve given us a five star review on iTunes, if you have a chance opportunity. Go look at our page, you’ll see all the places where we’re listed for Dave Rohrer with Northside metrics and met soldier with avalanche media. Thanks, guys. Bye
  • Microsoft Word – Speaker 1 Alrighty, Well you guys um. Speaker 1 Make sure that you’ve given us a five star review on iTunes. If you have a chance opportunity, go look at our page. You’ll see all the places where we’re listed for Dave War with North side metrics and metal with avalanche media. Thanks guys, bye.
  • Temi – Speaker 1 (16:37): Alrighty. Well you guys, um, make sure that, uh, you’ve given us a five star review on iTunes. If you have a chance opportunity, go look at our page. You’ll see all the places where we’re listed, uh, for Dave war with Northside metrics. I met salsa with avalanche media. Thanks guys. Bye.
  • Audext – Matt Siltala 00:16:37 – 00:16:55 already. Well, you guys make sure that you’ve given us a five star review on iTunes. If you have a chance opportunity, Go look at our page. You’ll see all the places where were listed for Dave War with North Side metrics and Met Social with avalanche media. Thanks, guys,
  • Sonix AI – [00:16:37] All righty. Well, you guys. Make sure that you’ve given us a five star review on iTunes, if you have a chance opportunity, go look at our page. You’ll see all the places where we’re listed for Daviau with North Side Metrics and Matsuzawa Avalanche Media. Thanks, guys. By.
  • Amazon Transcribe – N/A
  • Happy Scribe – N/A
  • Trint – [00:16:37] All righty. Well, you guys. Make sure that you’ve given us a five star review on items, if you have a chance, opportunity, go look at our page. You’ll see all the places where we’re listed for Day Raw with Northside metrics and mezzos with avalanched media. Thanks, guys. Bye.

Full Transcript of Episode

Matt Siltala: [00:00:00] Welcome to another exciting episode of the business of digital podcast, featuring your host, Matt silt and Dave roar. Hey guys, excited to have everyone join us on another one of these business of digital podcast episodes. And we are just going to jump right into it. This is one that, uh, that a lot of people ask Dave and us this question, first of all, Dave, how’s

Dave Rohrer: [00:00:26] it going?

Good ma make a squeaky chair.

Matt Siltala: [00:00:29] Hey, I don’t have

Dave Rohrer: [00:00:29] squeaky chair today. Yeah.

Matt Siltala: [00:00:31] I can’t make squeaky chair.

Dave Rohrer: [00:00:33] One less thing for me to edit, but also like your sound today is like ridiculously much better than mine and it makes me shameful.

Matt Siltala: [00:00:41] I’m actually coming to you guys straight out of a real live podcast studio, if you can believe that.

So, but, uh, what we’re going to talk about today is what is the best transcription service and. This would be, I guess it’s not just for podcasting. It could probably be for other stuff, right, Dave, or is it,

[00:01:00] Dave Rohrer: [00:01:00] you can use it, is it really? Yeah. For anything, um, any sort of video or audio file that you need or that you want to put up, like if you were creating YouTube videos and you want to have, um, closed caption and you want to be able to, you know, We will talk about that to a certain extent as I go through the reviews.

But yeah, that’s, that’s another thing. Anytime you have video or any sort of audio, or if you want to listen to, um, we’re just going to start getting right into it, all of the uses for it. Um, sales calls, customer service calls.

Matt Siltala: [00:01:34] Yeah. That’s what I was going to ask. Like, um, because I know that obviously the SEOs and us want to use this simply for.

The content we can get it put out there, you know, we’re there’s, these are audio files, but it does make it good to get this actual text that we’re talking about and whatever the topic may be. And so, um, I, that is exactly what I want you to jump into and like, what are all the uses for it? What is the best [00:02:00] transcription services that people can use?

I know that we have a, a document here full of. Just different ones in pricing. And I’m sure that, you know, depending on what we get into for everyone, you’re going to have this as a resource to available to everyone as well. And so I’m just, yeah, just kick us off

Dave Rohrer: [00:02:15] Dave. Yeah. If there was ever a time where you should actually look at the writeup, this is probably, um, longer than anything I’ve written in years.

Not even including the copy and paste where I actually copied. So you can actually compare yourself. The transcriptions from each of the services. So we look at, and, or I looked at

Matt Siltala: [00:02:39] just because Dave was crying to me about

Dave Rohrer: [00:02:41] it all week. One, two, well, one, two, three, four, five, six, seven, eight, nine, 10, 11. I looked at 11 services.

Wow. Um, and I only looked at AI services. So that’ll be the first caveat, as I will [00:03:00] say. I only looked at AI services and that is because if you look at this document, you will notice that it’s organized or sorted by the rating that I gave it. And I actually break down how I came up with that. It’s not just some, you know, Oh, I think it floods or it’s the best.

So it gets a nine. Um, I tried to be consistent across all four of the grading, um, items. But I also then brought in and looked at a couple services that do it with humans. And the one thing that you’ll jump out is if you really need a hundred percent accuracy or 99.9, nine, 9%, you go with a human that said, Matt don’t look and cheat.

How much do you think the average prices for a 20 minute? I think it was audio file transcribed by. Cumin. Boom.

Matt Siltala: [00:03:58] So a 20 minute file.

Dave Rohrer: [00:03:59] Yup.

[00:04:00] Matt Siltala: [00:04:00] Um, uh, like maybe 10 bucks.

Dave Rohrer: [00:04:06] Nope. Double that at least 20

Matt Siltala: [00:04:09] to 30.

Dave Rohrer: [00:04:09] Wow. Yeah. 20 to 30, depending on which of the four services you look at. Um, but then one of the things that’ll quickly jump out is.

Each of those services will specifically ask you how many speakers are there. If there’s more than two, it goes up per minute. Is your quality good? Five bucks? What’s that I

Matt Siltala: [00:04:32] said, or if we use our kids, we can pay him five bucks, you know?

Dave Rohrer: [00:04:34] Yeah. But you can also, you know, what’s how long has the turnaround going to be?

How good is the quality? Um,

Matt Siltala: [00:04:42] everybody don’t do that.

Dave Rohrer: [00:04:44] Yeah. So I looked at, um, Casey and I. I apologize, Cassie, Cassie, public Cassie, um, how Schlitt rev go transcript and GMR transcription services were the four that I just kind of found and poked around. [00:05:00] And most of these services, including them were ones that were listed multiple different places for the best service or my favorite service or something like that.

And then I just kinda did some additional searching on. Additional listing sites and just everything else trying to figure out which ones worked. Um, I did look at two software solutions in scribe, um, which has the loo lowest score possible just about, I could not get it to work, which is part of the problem.

Um, I got it to load and, um, I can pretty much do the review right now on it. I got it to load, but I had Java issues and. Something else was wrong. And I don’t know. It just like, it just wouldn’t do anything.

Matt Siltala: [00:05:43] Definitely have to say you pay what you, you get, what you pay for with that.

Dave Rohrer: [00:05:47] It was, it was a it’s $99 one time.

I do believe. And you know, if you’re going to do a lot of transcriptions, I could see that being a good solution. [00:06:00] I couldn’t get it to work though. Interesting. So, you know, I’m on a. I am on my trusty old windows computer. Um, and for some reason it w I had quick, I didn’t have quick time. I re-installed quick time.

It still wouldn’t work. I couldn’t get it to co I couldn’t get it to work. I couldn’t find anything out about it. I couldn’t figure it out. So maybe for those that have a Mac, it would work better. Uh, but for $99, you know, if you’re doing. 40 to 50, if you’re doing a weekly podcast and over two years, I mean, it comes out to be a buck.

Matt Siltala: [00:06:41] That’s a very good deal.

Dave Rohrer: [00:06:42] I mean, for a 20 minute episode, and that’s kind of what I looked at, and there’s a 14 day trial, which is what I used, but for a dollar to transcribe 20 minutes on average, you know, and then the same thing, if you have lots of different files of [00:07:00] customer service and other stuff, and, you know, imagine.

Three sales people a day or five customer service people a day. And you want to listen to all of their calls throughout the day for, you know, six hours a day, that’s 18 hours a day between three customer service people you want to transcribe that can get really expensive. So, um, the other solution that I used, um, that was, um, a download is descript.

Which Matt knows about because that’s what we use for our transcriptions. Um, and I will save you from having to dig through everything by telling you which one, you know, unless you want to cheat and go and actually look at the document of our reviews, but I will tell you which one was my favorite. I can tell you, you know, the fact that I still use transcript after doing this last couple of weeks probably tells you that it’s up there.

[00:08:00] Um, but yeah, a lot of the services, whether it was an AI or human edited, there’s lots of questions that would come up. It’s how many speakers do you have? Do you have accents was one interesting. Um, what is your budget? How long are your episodes? Because some of them. To try it, or just in general, they, there were certain files that it would work better or not like if you have a four hour, one of them has a, I think it’s otter.ai over if it’s, if you’re using the free one or even the paid one, it’ll only do up to four hour sessions.

So, you know, if you’re Joe Rogan, it could kind of do most of your shows. I don’t think many people are doing over a four hour. Anything, but I mean, think about all of the conferences now, they’re all online. If you wanted to take your video and transcribe it into a document, you know, if you do a webinar or a training and it’s two or three hours, you could still do it.

[00:09:00] Matt Siltala: [00:09:00] Yeah.

Dave Rohrer: [00:09:01] Um, but again, if you have a human do it at 20 to $30 for 20 minutes, you’re talking 60, $90 an hour. So you’re talking three to $400 to transcribe just one.

Um, what is your turnaround time? How often do you record, um, do you want to be able to do any editing? Do you care if it’s perfect? Honestly, I will tell you what we do. Not at least I don’t. Um, if you ever notice our names in our service and the transcription service, it’s always off and I just leave it there.

Cause I find it funny. Dave it’s spells it. R O a R every single time. The whole really? Yep. Just like a roar. Yup. Your name. It’s a mat. I think you shared that document with you. If you want to, you want her to giggle, scroll about halfway down where you’ll see the transcription sample [00:10:00] comparisons and it goes through and I broke out four different markers.

I did the intro, the outro because the intro and every single. Um, episode it’s it’s recorded, it’s prerecorded by Matt and we just apply that. So I was like, that is the most, Matt did it alone in one time. It’s always the same. It’s on the same on every show. So I thought that was an interesting one. Plus I knew it had both of our names on it.

Um, and that was really the only thing across them, other than some weird punctuation. So one of them got my name or a couple of them. Got it. As roar. R O a R a couple of them. Got it. As raw. Um, yours was a C, C O Tola C I O T O L a C O Tola um, one. Got it. Right. So Tala S I L T a [00:11:00] L a one. Got it. Is Silva. And another one.

Got it. As. Cecila S I L S I L a good, good, good. Oh yeah. Silt. Yeah. Salt, good effort. Good effort there, Tammy.

Matt Siltala: [00:11:15] Yeah, one that makes me laugh the most is the SIA Tola

Dave Rohrer: [00:11:20] yeah, I was like, what?

Matt Siltala: [00:11:22] But yeah. Dave, raw, raw,

Dave Rohrer: [00:11:24] raw. Yeah, that the Sonics one was the worst as far as the names. But the one thing that I came out of doing the, all of the digging and looking at all of these things, um, if I can like the next marker, I would talked about it.

It was, uh, SEO and API. Lots of the services got them wrong. So CMS, uh, SEO and API, those are all acronyms. Names and acronyms, most of the services get wrong. One of the things I looked at was the [00:12:00] feature and you can create your own word vocabulary. So for our last names for PPC, for SEO, for CMS, and of course, as I put this one through the transcription service later, I’m probably going to double check it because that’ll be funny as I just listed them all off.

Um, You can create that. So if you have your own jargon, your own words, your own names of, you know, if you’re a sports writer, if you’re, you have a podcast about sports, you have a podcast about legal things. And there’s very specific terms that you always want to make sure that, especially for legal that come up right, you can create your own vocabulary and use that, which I think is really cool.

Yeah, but some of them limit some of them, you know, depending on if you use the free $10 20, you know, enterprise, that size of the vocabulary will change. So again, back to the, how much are you going to transcribe? What is your budget? What matters to you? [00:13:00] Um, a lot of the scores that I gave for features, you know, I talk about it, but.

It’s a lot of what really matters and what doesn’t so ease of use is one of the things it’s, how easy is it to set up? How fast is it? Um, one of my problems I had with one really good service Otter was it was just slow and it is forever.

Matt Siltala: [00:13:22] And this kind of a thing becomes a big deal when you’re doing all that you’re doing, putting together an episode.

So

Dave Rohrer: [00:13:28] that’s well, so Otter was the slowest AI. If you use a human, it can be anywhere from three to five days is a rush. Some of them take one to two weeks. So the artery I in, in the, I want it now, world was slow. Like I had to wait and it emailed me like 20 or 30 minutes later, like, Oh, darn. Um, when I used a script, it literally is done within a minute, maybe two.

Matt Siltala: [00:13:58] That’s pretty insane. How fast it is. If you [00:14:00] think about it.

Dave Rohrer: [00:14:00] Yeah. And most of the services were like that. They were very, very quick. For some reason, Otter just was like, Hey, you want us to email you? Um, to be fair, I was also using the free service. So if there’s a queue and people using it, I would, I would put me at the back too.

Matt Siltala: [00:14:16] Yeah, that’s true.

Dave Rohrer: [00:14:19] Um, speaker handling one of the big things I saw, if you do, um, if you just have one person talking, it doesn’t matter. You know, the speaker handling is not something you care about, but if you have a salesperson and customer or a prospect or a customer and customer service, if you have two, three, four people on your podcast, if you have one or two people on your video or, you know, then it matters if you’re just doing a single video interview and it’s one person always, you know, talking that speaker handling doesn’t matter because some of the services honestly stink.

Is the nicest word I can say when it comes to [00:15:00] allowing you or making it easy for you to list who the speaker is. And if you look through the examples that I put in there, a lot of them won’t even say, well, just say, speaker one, speaker two is, is a nice thing, which is what, uh, the Microsoft does. A lot of them doesn’t even say who it is.

It just has, uh, some, some have a timestamp, some don’t. Was the other thing I noticed, like, Microsoft word is great, but it doesn’t have a timestamp, but it has the speakers. So at least, you know, who’s talking. But when I was comparing, trying to line up the same space, the same bit within. You know, a 20 minute or 16 minute file.

Most of them I could go and find it was like throughout the, the eight 45 Mark. I could just go and find it and then copy and paste it and try to compare apples to apples with Microsoft. I had to like, look for certain words because there is no timestamp. [00:16:00] So again, depending on what you’re going to use this for, you know, my ratings, I tried to break it out.

It will be helpful. So. I know what time are we at 16 minutes? We’re at 16 minutes of rambling about transcriptions. So for those that are really lazy and don’t want to look the two services I would highly recommend the most would be descript, which is what we use. And it’s a download or Otter. AI both have great speaker handling.

Both were quick, or I come home Otter wasn’t quick, but both were. Pretty good. Both were easy to use. Um, Otter AI was missing a couple of features that I would have liked, but, um, and, but like when the features is also exports, you might want to look if you’re doing close caption. Some of these services that I really dinged on features is because they [00:17:00] only will spit it out as a word file or a text file.

You can’t get it as. There’s like, Oh, was it descriptive? And, uh, Tammy, Tammy, um, who was it? Simon? AI, Simon says AI and one other one had like awesome list of like here’s 20 different ways. We can export this for you. Which one do you want it? It’s like, Oh, sweet. Um, another great service. If you use office three 65.

Go to the online version, go to like your browser and edge, open up Microsoft word three 65. And there’s actually a button that if you load up a file or if you speak into it, it’ll transcribe it for you for free. Um, that was my third best. It does lack some features. Like there is no integration that can’t integrate with zoom.

Like some of these services will integrate with zoom. So if you [00:18:00] have someone that wants to take wants notes from your meeting, boom, there you go. But you have to look at an Otter or think it was Simon. Simon says AI, and maybe it was Sonics that tie into your zoom or your, your Zapier. So if you drop a file and you know, Dropbox, if you’re grabs it, throws it right into it, you know, so there’s a lot of different things.

It just depends on what matters to you. Um, after that was also Timmy. I liked that service a lot. Um, it was just lacking some, some features, but it also starts to get really kind of expensive. It’s a $5 for a 20 minute file. Um, and then after that was odd text Sonic AI, Amazon actually has a transcribed service.

It’s called Amazon transcribe very original name. I’m happy scribe. Simon says AI and Trent, they start to get about three to $5 per 20 minute. And they [00:19:00] started to lack features or they started to really stink it. Um, speaker handling or timing, something like that. And not, so I just rated them lower. So that’s, that is the, I’m going to guess.

We, even, without the. Sample comparisons, the 2,003,000 word, like comparison right up and everything. And that’s the short version in 20 minutes, right on the dot almost 20 minutes. Um, so yeah. Think about what you want to pay. Think about how many speakers think about. What features matter to you? And I think, you know, there’s all the links to all these services, so you should be good to go, but if anyone ever asks us and I see it all the time, what transcription service should I use?

This is where I’ll be sending you there.

Matt Siltala: [00:19:52] It is very good guys. And so. I appreciate you taking the time, David, [00:20:00] you

Dave Rohrer: [00:20:00] know, David, just to read it.

Matt Siltala: [00:20:01] Yeah, exactly. So, all right. Like, uh, like Dave said, anything that we talked about and the full writeup, the full information is going to be on the page. So thank you guys for joining us as always a reminder.

We are on iTunes. So please go to iTunes and give us a five star rating. If you like the content that we’re bringing in, you want us to continue to bring this to you? That definitely helps. So thanks everybody for joining us. Uh, for Dave roar with Northside metrics, I met Scilla with avalanche media and, uh, thank you guys.

We’ll

Dave Rohrer: [00:20:33] talk to you on, I like how you annunciated your, your last name, that time

Matt Siltala: [00:20:38] we’re going to put these trends.

Dave Rohrer: [00:20:40] He wants to test it.

Matt Siltala: [00:20:41] Test.

Dave Rohrer: [00:20:41] Yes. Thanks all. Bye.

E172 – What is the Best Transcription Service? Hosts:



The Business of Digital © 2020