Transcription speech recognition, believe it or not, has been around since the 1980s.

And during that time, thanks to AI like Siri, Alexa, Google Assistant, IBM’s Watson and so on, it has improved by leaps and bounds in recognizing human speech.  Does transcription speech recognition software exist that can effectively replace humans in the transcription of dictations, one-on-one and multiple speaker interviews, seminars and teleconferences?

Let’s find out.

Best Transcription Speech Recognition Software 2019

DISCLAIMER:  At the time of writing all the software listed below are good for transcribing slow paced, American accent dictation recordings by a single speaker.  They work best when there is no background noise and you’re talking close to the microphone.  They struggle when there’s more than one speaker, background noise or music, or overlapping conversation.

Below is a list of the best transcription speech recognition software available today, in no particular order:

Dragon NaturallySpeaking 15

transcription speech recognition Dragon

Cost: $279.99

Operating System: Windows, Mac, iOS and Android

As one of the first transcription speech recognition software (released in the 90s), Dragon has been improved by its creators Nuance incrementally over the years to where its AI today is very good, but still not quite there yet.  Nuance claim it’s 99% accurate, but that’s not entirely true across the board.

It’s great for single speaker dictations, especially if you’re speaking slowly close to the microphone.  This means it’s an effective solution if you like taking notes on your digital recorder in a quiet place.  For example, a doctor taking medical notes, or a book author recording ideas.

Be warned though, your background needs to be absolutely silent!  Even the slightest background chatter or music, the software will struggle and errors will creep in.  In the same way it will have a problem with multiple speaker interviews.

Your first thought when you see the price of $279.99 is that for the individual version, it is expensive.  But this is more than just a transcription speech recognition software.  It’s a learning AI.  So, over time it learns to recognize and better understand your voice and accent.  You can use it to open your email, dictate the text and send.  You can open other projects and also launch applications on your computer.

Google Docs Voice Typing

transcription speech recognition Google Docs

Cost: FREE

Operating system: Android, Chrome OS, iOS and KaiOS (Jio Phone and Nokia 8110 4G)

Strictly speaking, this is not a transcription speech recognition software.  It’s a voice dictation software.  Or as Google itself calls it “voice typing”.  So, if you’ve recorded your interviews in MP3 or WAV or any other audio format, you won’t be able to run it through this as you would with Dragon or other services like ScriptoSphere, for example.

So, if what you want is a free service purely for converting your dictation to text, this is great.  It is surprisingly accurate, but you may have to clean up (proofread) the document for accuracy.  Or we can do it for you.

Braina Speech Recognition Software

Cost: $49 for 1 Year and $139 (or $239) for the Lifetime Version

Operating System: Windows 10/8.1/8/7/Vista/XP. Android and iOS App.

Yet to be officially confirmed, the “world’s best speech recognition software”, Braina also claims it allows you to accurately and easily dictate speech to text in 100 languages.  It also claims you can use it in a way similar to Dragon to control your computer to open programs and websites.

They also offer an Android or iOS app that can turn your smartphone into an external wireless microphone over a WiFi network.  Which we have to admit, is quite cool.

While Braina have built a large vocabulary into the software, it still comes up short if your audio quality isn’t the best, much like other transcription speech recognition software featured in this post.

TranscribeMe

Cost: $0.10 Per Minute for Transcription Speech Recognition

Operating System: PC, Mac, Mobile

This transcription speech recognition service costs $0.10 per minute.  Making it comparable to similar online services.  However, just like the others, its accuracy takes a hit the moment you introduce unclear audio quality to it.

Transcribe Wreally

Web Based Transcription Playback Software

Cost: $20 Per Year for Transcription Software

$0.10 Per Minute for Transcription Speech Recognition

Operating System: PC, Mac, Mobile

Another automatic transcription speech recognition service that demands only the best audio quality.  Anything less than perfect and it struggles, much like other software we’ve reviewed in this post.

You can however slow down the text and read it aloud yourself, which can bring the accuracy up.  If you have enough time.

You could also use it as a transcription playback software (like Express Scribe) by playing the audio file and using playback controls.

The second option will create a rough transcript, which you will have to format, clean up and proofread yourself.  Or use our proofreading services.

Amazon Transcribe

transcription speech recognition Amazon

Cost: $0.0004 Per Second for Transcription Speech Recognition

Operating System: PC, Mac, Mobile

This service from Amazon is not only suitable for developers who want to add transcription speech recognition to their apps.  But Amazon Transcribe can also be used to convert audio and video files to text.  You can transcribe up to 60 minutes a month for free for 1 year.  After that it’s $0.0004 per second, which means automatic transcription of 1 hour of audio i.e. 60 minutes, or 3,600 seconds will cost you $1.44.  However, that will provide a rough transcript which, again, you will have to proofread and format yourself.

Trint

Cost: £13.20 Pay Per Hour of Upload (PAYG)

Subscription: £36/Month and £100/Month

Operating System: PC, Mac, Mobile

A transcription speech recognition software that lets you get automatic transcripts of audio and video files.  There’s also a quick search and edit feature available that comes in handy to improve accuracy.  Their pricing is a bit high compared to other services featured here.  And again, just like the others, accuracy takes a hit as audio quality goes down.

IBM Watson Speech to Text

transcription speech recognition IBM-Watson

Cost: Free for up to 100 audio minutes per month

Contact for Standard and Premium Pricing

Operating System: PC, Mac, Mobile

This service can be tried for free for up to 100 minutes of audio per month on their Lite plan.  Then they have standard and premium pricing packages for real time transcription.  IBM claim their software can transcribe even bad quality audio with high accuracy.  But this has not been the case when we tried it.  Similar to Dragon, you have to spend time training the software.  Accuracy-wise we’d say it’s fairly similar to Dragon, but in some cases may be slightly better.

Temi

Cost: $0.10 Per Minute for Transcription Speech Recognition

Operating System: PC, Mac, Mobile

Their automated transcription speech recognition software gives you a transcript in under 5 minutes.  What’s cool is they have an app that lets you record audio, and then edit and share it.  It’s available for both iOS and Android.  The cost of their service is $0.10 per minute for automated transcription, which is similar to most of the services listed here.

Also, much like other automated transcription speech recognition services featured in this post, audio standard determines final transcript quality.  You also have the option to receive transcripts via email.

ScriptoSphere

transcription speech recognition ScriptoSphere

Cost:

Transcription Speech Recognition: $0.10 Per Minute

Human Transcription: Starting $1.20 Per Minute

See our Transcription Rates

Operating System: PC, Mac, Mobile

Best for the last? 😀

Our very own transcription speech recognition service starts at $0.10 per minute.  With over 15 years of experience in audio transcription, we have trained our software to learn from various types of English accents and document formats.  While we don’t offer automated transcription of different languages yet, watch this space as we may introduce it soon.

As we’ve already pointed out, voice recognition AI hasn’t come that far that it can transcribe your audio/video files accurately regardless of the audio quality.

The difference with our service, though, is that we also provide proofreading of machine transcripts.  So, our experienced quality analysts will format, label speakers and time stamp difficult words for you, while thoroughly proofreading the whole document.  We’ll also do grammar and spelling check for you.

For more info on rates, contact us.

Ways to Save Time and Money with Automated Transcription

We’ve done several posts on our blog giving you unbiased, factual insider tips and tricks to save money on transcription, like:

5 Ways to Pay Lesser for Interview Transcription Services

5 Ways to Save Time and Money with Academic Transcription

If you don’t have the time or the energy to click on the links above or read those posts in their entirety, or even skim through, let us give you a quick list of to-dos to save time and money with transcription speech recognition software.

  • Record in a Quiet Place.
  • Pay Attention to Recording Quality.
  • Avoid Multiple Speakers in one Interview (If you can)
  • Voice Recording App for your Smartphone.
  • Invest in a High Quality Digital Recorder (If you do this often)
  • Plan Ahead to Avoid Paying Extra.
  • Communicate Your Needs Clearly.
  • Don’t try and transcribe interviews yourself. (It takes a lot of time)

If You Already Have Voice Recognition Transcripts

If you have paid for or run your interviews through a transcription speech recognition software already, but don’t have time to format or proofread them.  Don’t worry.  There are proofreading services online like ours that can save you time and money.

Can Artificial Intelligence Replace Humans?

As a provider of transcription services, we can tell you that transcription speech recognition software has been at our disposal for a long time now.  We’ve gone through all iterations and different versions and evolutions trying to find the one that can provide accurate transcription.  And save us the hassle of finding, testing and hiring transcribers.  And save us the money to pay their salaries!  We are still searching.

Furthermore, they all claim to catch all words accurately.  But that’s only true if you speak very slow, like speaking to a child.  And in a quiet environment.  And as close to the microphone as possible.

Some of the biggest stumbling blocks for transcription speech recognition are:

  • Understanding Fast Speech.
  • Differentiating between Multiple Speakers.
  • Ignoring Background Noise.
  • Deciphering Foreign Accents.

Oftentimes they do struggle even with slow pace of speaking.

Even if any of the transcription speech recognition software that we’ve listed earlier in the post understood slow speakers perfectly, you’d still require human editors to refine the rough transcript.

Example of a Transcription Speech Recognition Transcript

Below is an example of how a machine understands people talking.  This is a quick transcription of a small clip from the incredibly popular Joe Rogan Experience Podcast Series.  So, when we talked about a “rough transcript” earlier, we meant run on sentences, no punctuation, no speaker labeling, no grammar etc.

Is telling you guys before issue today were my credit card got robbed and you know whatever card fraud some to hold my number which is really that’s the one thing that people worry about the most about shopping online right because I could, happen everywhere but you think that if there’s anything that is change malls more than anything it’s got to be the ability to just shop online default Amazon that the fact that they figure that out when Amazon came out it was just a bookstore with who the focus of our books online at school to the bookstore is just a stupid business meanwhile the guy has more money than any human on the planet that’s related to investors get divorced I heard last night that there’s a rule in West Washington that everything that you make as a couple you to split your for 25 years else Arnie a little military law is just 50-50 will under marriages 20 years or 10 years or more is 50-50 in that clues downstream income will go from anything you did that you’re still getting paid for say 10 years ago you wrote a book or whatever well Mrs. basals is again a he’s worth 100+ billion dollar and 37 while so she’s gonna get half of that ligament because it’s a lot of stock the center to just be friends, though they better if she should be supervised usual user recorder actually I were on July were on the richest people in the are all these bookstores where business or some irony that Amazon I want to start opening physical brick-and-mortar stores that is, in my life but it does make sense though because there are some brick-and-mortar stores and think of as a conqueror think you just want to take over everything will wildly by whole foods and I was out of the supermarket business I can talk this up to well automated I can order whole foods delivered to your house I think that’s available yet but I figured is right is a state now most grocery stores even like all the routes have home delivery within two hours to compete with Amazon prime you are and if they don’t do that you can is where I want to pick it up to John I went to pavilions in the day met a bunch of delivery trucks parked in the parking mom is what will get this is cool delivery trucks have to buy one rate that we can walk in there’s no cashiers they just trust that everyone’s not stealing and I think they just are the comfortable with a certain amount of theft that’s interesting with the left hand was doing but it’s him the online purchasing like that is got a put that must be putting the most this was like the most impact on stores brick-and-mortar stores and ability to purchase online can be devastating from Dorian up in Santa Barbara I live now just writing up State Street this morning at the end of the ride units like the third to be a quarter the stores are closed and business empty and you state streets and Barbara this was like that happen in place to be and I and the real estate minimizes I will see that started at zero $22,000 a month to lease okay so you have to have a retail outlet that’s really turning over the customers and there’s a lot that just can’t do that became an antique store that’s going to do that are a little bit next-door

As you can see in the sample above, the transcription speech recognition AI does fine as long as the speaker speaks slowly, into the microphone, and there’s no overlapping conversation.  Things go south very quickly the moment Joe Rogan picks up his pace or the guests speak at the same time.

Which is why transcription speech recognition is great for single speaker slow dictations, but may not be suitable for 2 or more speaker interviews.

Is Transcription Speech Recognition Really Low Cost?

Sure, transcribing a 60 minute file is going to cost you only about $6 at $0.10 per minute.  But as you can see in the rough transcript above, unless you’re experienced and really quick on a computer, it will take you time to clean that up.

Now, time = money.  And cleaning up a rough transcript of a 1 hour interview can take you anywhere between 6-8 hours.  Depending on how much you earn per hour, we can guesstimate it’s probably an average of $12 to $15 per hour, maybe more.  Unless you’re on minimum wage.

So, at an average of $15 per hour, you’ll lose around 8 x 15 = $120 by doing it yourself, possibly.

Human Transcription: Cheaper, Safer, Better

If you’re going to spend time cleaning up a machine transcript, why not just find a transcription service online and have your peace of mind?  At low transcription rates of $1.20 per audio minute, a 1 hour interview is only going to cost you 60 x $1.20 = $72.  And for that you get a well formatted, speaker labeled, time stamped, proofed, spell-checked, grammar-refined, almost 99.9% accurate transcript.

It’s a no brainer.

While transcription speech recognition is getting there, it can’t beat humans when it comes to transcription yet.  And it most probably never will, unless they refine and evolve AI to the point where it functions equal to or better than a human brain.  Which, for many reasons including safety, may never happen.

AI is Dangerous. Haven’t you seen The Terminator?

transcription speech recognition AI vs human

From transcription speech recognition to end of the world.  WHAT!?? 😀

All in good fun.

It looks funny because it seems like needless scaremongering.

However, there is a potential danger.  The problem we see with a sentient AI is it could ignore laws around confidentiality.  Which means even if the company you’ve chosen may sign an NDA, your data still goes through a machine database and may get stored there permanently.  To be used for learning, or maybe more.  This is one of the reasons why Elon Musk has warned against AI.

So in the immediate future, rely on humans who you can trust, who are bound by legal confidentiality agreements and still respect international laws.