I wrote a few days ago about how I’d setup a dedicated email account for my Newsletter addiction. This was to stop me getting distracted and falling down a rabbit hole of learning. Something I really don’t need at 10:30am on a Tuesday morning when what I should be doing is making my 3rd cup of tea of the day, and getting on with some actual work.
As a British man, there is a direct correlation between my output and the number of cups of tea I drink.
Anyway, for the past year or so, I’ve been experimenting with using AI voices to read long articles and emails to me – I’ve found that I retain a lot more information when I’m both reading and listening to text. Something to do with engaging multiple senses (makes sense).
Whilst AI voiceovers have been bloody good for a while, they haven’t been perfect – until now.
Speechify
I first started out with Speechify, an app that boasts, Gwyneth Paltrow, Snoop Dogg and Mr. Beast as voiceover artists. All of which are nothing more than novelty voices (IMO).
Speechify gives you many voices to choose from and their integration with Chrome, Mac and iOS is good, if not slightly buggy at times. The whole service is a bit pricey if you go for their Pro plan, which unlocks everything – $139 a year.
Generally, the voices are good, but they still fall into the “uncanny valley” area. You can tell they are AI, and after a couple of weeks of use, I stopped using the service (I was still on their free plan).
Apple
Then, a kind soul on Twitter alerted me to the native built-in voice options on Mac. Think Siri and several additional voices. However, after a while, having Siri or any of their counterparts reading to you becomes slightly unsettling.
This was getting on for six months ago now, when I decided to put my AI voiceover dream on the back burner until the tech caught up with my expectations.
ElevenLabs
During this temporary pause I’d been aware of the AI communication startup ElevenLabs, and their unmatched AI voices. They have trumped Speechify and Apple, but with great quality comes a big price tag, or in this case a low token limit, meaning few text to voice minutes.
ElevenLabs free plan offers 10 minutes of text to voice per month, which I could hit in a single listening session. Their starter plan isn’t much better giving you 30 minutes for $5 a month. So, ElevenLabs has been off the table.
As a general note… as LLMs (Large Language Models) become more advanced, the cost of processing tokens tends to decrease. This makes AI tasks, like an AI voice reading me my newsletters, more affordable and accessible for everyday use.
On the 21st August, Elevenlabs rolled out their Text Reader app worldwide FOR FREE. Available on both iOS and Android and in 32 languages. Simply put, the app allows you to paste/write text, import a URL, upload a file or scan a page and within seconds you can have the text read back to you in perfect voice.
Not to be outdone by Speechify’s celebrity roster, ElevenLabs has licensed the voices of some proper Hollywood royalty: Laurence Olivier, James Dean, Burt Reynolds, and Judy Garland. I don’t know who would use these voices mind – probably more for the marketing department at ElevenLabs than anything.
Celebrities aside, ElevenLabs have a huge roster of AI voices to choose from in all sorts of accents and tones.
Over the years, I’ve grown rather fond of those softly-spoken American male voices you often get on audiobooks. You know the ones – think Wil Wheaton or Dylan Baker. Proper easy on the ears.
His Name Is Brian Overturf
So with my preferred styled already dialled in – I’ve opted for Brian Overturf the professional narrative story voice. Brian’s dulcet tones are the perfect accompaniment for my daily newsletter sessions.
When I say these AI voices are now perfect, I mean it. With the exception of the AI getting stuck, if you asked Joe Public to pick out the AI voice in a lineup, they’d be guessing.
It’s a bit frightening that in a few years (maybe months even), the person I converse with most each day, could be Brian Overturf. Love you, buddy!
I’m thinking Jarvis / Iron Man sort of vibes.
Text Reader is a superb app, but it is currently has no desktop equivalent (unless you pay), which makes it a pain to load up text when you do most of your reading at your desk. ElevenLabs have also stipulated that the app won’t be 100% free forever.
I’m far from being adverse to paying for apps, but I need nearly unlimited use per month to warrant any cost.
Long gone are the day’s of Microsoft Sam, we are now in unchartered territory where voice manipulation could pose serious security issues. That, though is a concern for another article that won’t be written by me.
I’m an AI optimist and I’m excited for all the future lessons Brian Overturf will be teaching me.
Update: Saturday, 7 Sep 2024 at 10:06
I found an app called AudioBuddy which is a free MacApp that turns text to speech using OpenAI (you need to add your own API key). I haven’t tried it yet as I’m happy with ElevenLabs but it may be one for the future.
