A Survey of Voice-to-Text Options on the Mac, iPad, and iPhone

A lot of folks have been asking me what I’m doing about voice-to-text dictation now that Nuance has officially pulled out of the Mac and appears to have relatively abandoned iOS. If this is news to you, Nuance, the makers of Dragon Professional Individual for Mac, speech-to-text software, announced back in October that they would no longer support the Mac. It was disappointing news. Nuance has always been the leader in voice-to-speech dictation, and I have been a customer of theirs for years on both the Windows PC and Mac platforms. 

If you already bought Dragon for Mac, you will find that it still works. I have no idea how long that will continue. If I had to bet a nickel, I would bet installation of macOS Catalina as the day that Dragon dies on my Mac. But at least for now, you still have working software.

Over on the iPad and iPhone, the story gets even weirder. Nuance says they are still supporting those platforms. Their product Dragon Anywhere is, in my opinion, the best dictation solution on the iPhone and iPad. For years I gladly paid $14.99 a month to have that service because it was so useful on the iPad. Specifically, I had a workflow where I would look at a PDF and then dictate comments about it at the same time using the Dragon Anywhere software. It worked great, particularly with split screen.

That all came to a screeching halt when Dragon Anywhere just stopped working on my iPad Pro. The software went from being the best in class to complete garbage. It would drop entire sentences and generally not work. I looked into this a bit further and discovered that the problem is with the iPad Pro, where I did all of my mobile dictation with Dragon Anywhere. (I find Siri dictation just fine for short emails and text messages. Dragon Anywhere is what I need for the big jobs, which all occur on the iPad.)

For whatever reason, however, Dragon Anywhere no longer works on the iPad Pro. This is a known issue and has been for months. Dragon Anywhere is getting destroyed in the reviews in the App Store because of this failure and has, at least to my knowledge, not made any public comment or commitment to getting its software working on iPad Pro. Dragon Anywhere does, however, work on the iPad mini and iPhone just fine. Like I said, weird.

If Nuance has abandoned the Mac and seems only very slightly interested in the iPad and iPhone, what are we to do? The answer is … complicated. 

Solving This Problem on the Macintosh

There is no clear successor to Dragon Professional Individual for Mac. Siri dictation has made great strides recently, but it still has a long ways to go. Two critical elements for any voice-to-text dictation system are the ability to keep an open mic and the ability to store custom dictionary values. If I have a client with a unique name, I should be able to teach that name to the software so I can dictate it without having to go back later and fix it. If the software doesn’t support that, I have to use some silly name, like “Tiger”, and then go back later to do a search and replace on that unique name. That’s more work. I don’t like more work.

Siri dictation on the Mac ticks off only one of these boxes. It can indeed leave an open mic, but it has no custom dictionary. Moreover, it just isn’t as good as Dragon for the Mac. It’s close. Indeed, it is much closer than it used to be. But if you are used to Dragon, you will notice a difference. I demonstrate this in the attached video.

Another solution would be to install a local instance of Windows and then run the Dragon for Windows on your Mac. Don’t laugh. I have heard from many listeners and readers who have done this. Frankly, I don’t blame them. If you spend a lot of time dictating to your Mac to get your work done, you need the best possible tool, and, as the attached videos demonstrate, Siri dictation just isn’t there yet.

You could also hire it out. There are a lot of good solutions now with web-based dictation. Some use robots, like Temi, and others use humans, like GoTranscript. The robot-based systems are not, in my experience, any better than Siri dictation and often worse. The human-based services are deadly accurate and can be quite expensive, but depending on how you charge your time, it may be the best solution for you.

The point is that I’m just not happy with any of the alternatives to Dragon Professional Individual for Mac. I have been whistling past the graveyard for the past several months as Dragon continues to work on my Mac. One day in the future, probably September with the release of Catalina, that will end. At that point, I may seriously consider doing a Windows install just to have the Nuance software on my desktop. Another alternative I will seriously consider at that time, assuming Nuance hasn’t completely abandon Dragon Anywhere, is to dictate on the iPad and iPhone. I find Dragon Anywhere almost (but not quite) as good as Dragon for Mac.

Solving This Problem on the iPad and iPhone

Apple has made continued strides with Siri dictation on the iPhone and iPad. I find it very useful. The trick is to be precise when you speak to it and not mumble your words. You also have to understand the limitations of the application. There is a timer. After about four sentences, it’s going to end. If it ends midsentence, you are going to have a grammar crash on your hands and have to spend time fixing things later. You always have to be aware when you hit that third sentence and stop the dictation, and then start again. The Drafts application gets around this, sort of, but even it sometimes stumbles when the dictation timer resets during a dictation session.

If you are voice-to-text curious, Siri dictation on your iPad and iPhone is a great place to start. You can use it anywhere that you are using the keyboard. Just tap the microphone button and start talking.

If you want more serious dictation tools, like dictation without a timer and a custom dictionary, Dragon Anywhere is probably still best, provided that you are only using it on an iPhone or a non-iPad Pro. But it is hard for me to recommend this product at this point because it seems like the developer isn’t entirely convinced it wants to continue to support it. Just like on the Mac, it is still working for me (sort of) and I’m still using it.

Also, just like the Mac, there are third-party services that can transcribe your voice via robot or human for a fee with varying degrees of success, but as I explained earlier, quality can get expensive.

A Murky Ending

The reason I have delayed publishing this article is that I kept thinking I would find the magic solution and, frankly, it doesn’t exist. Everything I have discussed in this article requires some sort of compromise, whether it be money, quality, convenience (like installing Windows on your Mac), or all of the above.

Ultimately, the solution to this problem needs to come from Apple. Specifically, Siri dictation needs to be just as good as Dragon Professional Individu
al for Mac. It needs to support a custom dictionary, and it needs to be as reliable with its dictation engine. I understand this doesn’t happen overnight. I also suspect Apple is spending quite a bit of money to try and bridge that gap.

Meanwhile, however, there is a gap. Switching to Siri dictation is not going to work as well as Dragon for Mac or Dragon Anywhere. So for the foreseeable future, the battle carries on.

I have made a video to go along with this article comparing Dragon Anywhere and Dragon for Mac with Siri dictation on both platforms. Watch the video to get a better idea and hear some more of my thoughts on the state of voice-to-text dictation.

Dragon Professional Individual for Mac, v6 Review


There’s no secret that I’m a voice dictation fan. I know a lot of folks that enjoy fancy pens and artisan notebooks but for me words (particularly first draft words) are more something that I want to get out of my system than something to lavish upon. I started using dictation tools about 20 years ago, when they were pretty crappy.

Nowadays, however, our computers are a lot faster than those of 20 years ago and voice dictation software is quite a bit smarter. One of the leaders in this space is (and always has been) Nuance and its Dragon Professional for Mac dictation software.

A Brief History Lesson about Dragon and the Mac

For a long time, the Mac was a wasteland for dictation software. In 2008, a product called “MacSpeech Dictate” showed up and it was workable, so long as you had never used the superior Dragon Dictate on PC. (I used to run a virtual PC on my Mac largely for the purpose of using Dragon Dictate for PC.) Then MacSpeech licensed the Dragon dictation engine and things got better. A year later, Nuance bought MacSpeech Dictate outright and turned it into Dragon Professional for Mac. Since then Dragon Professional for Mac has been the gold standard for dictation on the Mac. The built-in macOS dictation isn’t bad but also can’t keep up with Dragon Professional for Mac on accuracy or features.

The Skinny on Version 6

Every year or two, Nuance releases an update to Dragon for Mac. Version 6 just landed and I’ve been using the heck out of it. The short version of this story is that dictation and usability improvements make Dragon Professional for Mac version 6 both a great product to get in on if you want to get serious about dictation and a worthy upgrade for existing users. Here come the details…

Easier On-Ramp

One of the reasons a lot of people don’t get very far with dictation is the training process. It takes time to get your microphone set properly and then train the application to understand your voice and speech peculiarities. This new version  does away with much of the previously required training and instead does a better job of learning through your use of the product.

The app is also better at distinguishing your words from background noise. That makes the app more forgiving. I normally dictate using my high quality podcast microphone. However, for the past few weeks, I’ve been using my Mac’s internal microphone to dictate words (including these). There is a small accuracy hit but it’s really not that bad. A few years ago, the application was unusable with built-in microphones. Now I think you could pull this off, even in an environment with low background noise.

Dictation Speed and Accuracy

Every version of Dragon Dictate boasts improved accuracy. Nuance reports this update improves accuracy by 15%. So long as I dictate complete sentences and don’t try to change course in the middle of a thought, I was already getting very high accuracy with version 5. Indeed, this is the secret to all dictation. If you give the application clear enough enunciation and context, Dragon can be extremely accurate. With a 1,000 word dictation, I normally have 5-10 corrections. Let’s call it 99%. With version 6, I’m still getting excellent accuracy. I ran a few tests and I’m getting about the same numbers with version 6.

The difference, however, with version 6 is dictation speed. Put simply, the words are showing up on the screen faster, making dictation easier. I like that.

There are a couple reasons for this. One is that the engine is just faster. Another is some underlying technology improvements with the way Dragon views words on your screen. Up until now, typing and dictating at the same time with Dragon on a Mac was the dictation equivalent of crossing the streams. That’s not true with Version 6. Through clever use of Apple’s accessibility API, Dragon can now monitor text and edits you make on screen while you dictate. This only works in applications that support the accessibility API but Scrivener, TextEdit, and Pages are among them. New apps are getting added as Nuance verifies their accessibility support.

User Interface

While the underlying engine of Dragon Dictate for Mac has been on par with its PC counterpart for years, the Mac version’s user interface has been playing catch up with the much more mature PC version. Nuance made big strides with version 6 with revamped status and correction windows that feel less … well … goofy. They also cleaned up the way the windows display commands and the preferences selection screen. These changes are a welcome face lift but they were also designed to reduce the number of clicks and scrolls required to get things done. 


Transcription

Transcription, the act of extracting text from a pre-recorded voice file, used to be a separate product but got added to Dragon Dictate as a built-in feature a few versions ago. However, it always felt a bit like it was bolted on to Dragon Dictate. The new version fixes this. Transcription is now fully integrated into the application. Moreover, they’ve removed the need for training. Dragon instead trains its transcription engine with the first 90-seconds of your first transcription file. They’ve also added a batch transcription tool. I often record several small recordings and transcribing them all at once is a nice improvement.

The Ecosystem

In the last year Nuance has improved the ecosystem around their dictation software. I also use Dragon Anywhere for dictation on iOS. With the latest version of Dragon Dictate for Mac, custom vocabulary words are now shared between Mac and iOS devices. If I add a new word on my iPad and then find myself dictating it on the Mac a few days later, it just works.

I use Dragon Professional for Mac every day. I think a lot of people that gave up on dictation 10 years ago are missing out. If that’s you, check out the new Dragon.

Dragon for Mac Version 6 Coming Soon

This week Nuance announced the imminent release of Dragon for Mac, Version 6. I spoke with Nuance and this new version takes advantage of several improved dictation technologies.

Deep Learning

Nuance has always been able to server-based algorithms to improve dictation accuracy but this new version will be the first time they can embed learning on a user’s computer, allowing them to improve their own language and acoustic model. This, and other improvements, adds up to less required training and a reported 24% improvement in accuracy. (I’m looking forward to putting that to the test.)

Improved Transcription

The demonstration I saw showed significant improvement in transcription of existing audio. Not only is the transcription better, it’s also much easier to train and operate. I particularly like the new batch mode, that lets you transcribe batches of audio files in one go.

Improved Text Control

Mixing typing and transcription has always been rough going on the Mac. With Dragon 6 for Mac, you’ll be able to dictate in supporting apps and type at the same time without the wheels falling off. They are still working on the list of supporting apps for launch but Scrivener is already one of them.

The new version ships (digitally) on September 1 and there will be physical product shipments by mid-September.

Dragon Anywhere Initial Impressions

Nuance has made some big changes in the last few months to its Dragon Dictate applications. They released a new version for the Mac, which I’ve been using and will be publishing a review about shortly. They’ve also changed their iOS strategy and released a new application, Dragon Anywhere (iOS App Store).

Dragon Anywhere is a dedicated dictation application for the iPhone and iPad. You may wonder why you would need such an application when Siri dictation is already pretty good. The problem with Siri dictation, however, is that it has a short fuse. At most you can get two or three sentences out before it stops. With Dragon Anywhere, there is no such limitation. You can leave the application open, turn on the microphone, and start yakking away. I’m doing that right now as I dictate this article. There are additional features, like custom words that synchronize between your devices.

Dragon Anywhere also brings a lot of PC and Mac functions to the iPad and iPhone that you don’t have with Siri. For instance I can select and edit text, use phrases like “scratch that”, and other dictation power user tricks.

I’ve only been at it a few days now but already I really enjoy using it on the iPad Pro. I’ve been opening it up as a split screen application with research on the left side of the screen and dictation on the right. It works great. Nuance claims 99% accuracy. I’m not getting it that accurate but it is more accurate than Siri dictation and adding my own words makes a bigger difference than you’d think. As an example, the screenshot below is the actual dictation without modification and you will see there are some errors. (Click on the screenshot for a bigger view.)



I can already see that the sharing tool needs some work. It is overly Microsoft Word focused. There is no one-tap the button to simply save text to the clipboard. I’ve been getting over that by using the verbal commands “select all” and “copy that”.

It still requires that you have an active Internet connection and it’s pricier than most iOS apps. It’s a subscription-based model that runs $15 a month or $150 year. While it may not be worth it if you only use it occasionally, if you dictate a lot, it’s at least worth consideration. (They have a free one-week trial.) I still have a few days left on my trial but I expect I probably will subscribe because of the amount dictation I do. Either way, expect a more thorough review of this product in a few weeks after I’ve had a bit more time to kick the tires.

Dragon Dictate 2.5 and my iPhone

In case you missed it, Dragon Dictate hit version 2.5 this week. As a long time user of the Windows version (Dragon was the only thing tying me to a Windows machine), I’m thrilled to see Nuance giving my favorite platform the love and attention it deserves.

Of note with release 2.5 is the ability to dictate to your iPhone (using this app) and have the words appear on your Mac. While you are giving up “hands free” it is really nice being able to tap a button on your iPhone, start talking, and see the words appear on your Mac. The developer explains that since the connection is Wi-Fi, it is better quality and less prone to error than Bluetooth. I never had many problems connecting to Dragon with Bluetooth so I’m not sure about that.

Regardless, if you are a Dragon Dictate user, upgrade and try dictating with you iPhone. You’ll get hooked.

Dragon Voice and OmniFocus iPhone Workflow

I’ve been using Nuance’s Dragon Dictation on my iPhone now a few weeks. One of my favorite uses for it is dictating task item into OmniFocus. Dropping new items in the inbox is simple.

 

Dictate the task into Dragon

 

Copy the task to clipboard


Open and paste the task into OmniFocus.

 

The Task Is In


Note that OmniFocus allows you paste items in the inbox even while the database is still updating. This has largely replaced my prior use of the Note2Self app where I would dictate new tasks and then process them after receiving the audio files. I know there are also other options, like ReQall but I prefer the above method.

Also note Nuance has updated the Dictate app allowing you to opt in (or opt out) of sharing your contact names. I know a lot of people felt this was a security risk. I don’t think it is that big a deal and uploaded mine.