Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Natural text to speech free?

S

stanleystf

macrumors regular
Original poster
Apr 24, 2022
213
36
Is there a MAC app that reads text aloud and allows you to convert text to mp3? Please note that I don't want to use Mac's built-in features. I'm looking for an app like this that's hopefully free:

apps.apple.com

‎Text to Speech-NaturalReader

‎With the Text Reader app turn your text into lifelike synthesized speech. Enable fluid, natural-sounding text to speech that matches the intonation and emotion of human voices. Text to Speech can export your tuned audio files effortlessly, thereby lets you create a high-quality audio file and...
apps.apple.com
 
Please describe why you won't use the built-in features. If we know why it won't meet your requirements, then we can make better suggestions.

In particular, the 'say' command in Terminal has many options for voice, quality, rate, audio format, etc. See 'man say' for details.

Also remember that any Terminal command can be placed into a Shortcut or a scripted droplet, and have files of text dropped onto it for processing. That is, the "type a command" part can be automated and used repeatedly.
 
  • Like
Reactions: schnaps, Digital Dude, stanleystf and 1 other person
chown33 said:
Please describe why you won't use the built-in features. If we know why it won't meet your requirements, then we can make better suggestions.

In particular, the 'say' command in Terminal has many options for voice, quality, rate, audio format, etc. See 'man say' for details.

Also remember that any Terminal command can be placed into a Shortcut or a scripted droplet, and have files of text dropped onto it for processing. That is, the "type a command" part can be automated and used repeatedly.
Click to expand...
I can't use the built-in features to create MP3 files out of specific texts. Also, when I use built-in features to read aloud texts in other apps, say Microsoft Word, it sounds unnatural and robotic.
This app, for instance, sounds a lot more natural than the Mac's built in voice.

apps.apple.com

‎Text to Speech-NaturalReader

‎With the Text Reader app turn your text into lifelike synthesized speech. Enable fluid, natural-sounding text to speech that matches the intonation and emotion of human voices. Text to Speech can export your tuned audio files effortlessly, thereby lets you create a high-quality audio file and...
apps.apple.com

I also downloaded extra enhanced and premium voices for Mac in the Spoken Content but still it sounds unnatural.
 
You've left out a lot of details of what you tried. Suppose someone else wanted to try and replicate what you're experiencing. Please try giving enough detail for someone to do that.

Some example details (not an exhaustive list):
What built-in voice did you use?
What downloaded voices?
What built-in apps did you use to read text aloud?
What OS version was being used?
 
  • Like
Reactions: stanleystf
Screenshot 2024-10-27 at 7.37.38 PM.png

I use MacOs Sequoia 15.

I used Terminal to convert a text file to an MP3 file and it sounded very robotic.

In Microsoft Word, I go to 'Review' and click on Read Aloud. Again, when it reads text it sounds robotic.

My objective is to covert a text file into an audio file and listen to it while I'm driving.
 
stanleystf said:
View attachment 2442591
I use MacOs Sequoia 15.

I used Terminal to convert a text file to an MP3 file and it sounded very robotic.

In Microsoft Word, I go to 'Review' and click on Read Aloud. Again, when it reads text it sounds robotic.

My objective is to covert a text file into an audio file and listen to it while I'm driving.
Click to expand...
Thanks for the additional info.

Please post the exact command line you used in Terminal.

Did you try any apps other than Word? For example, I think TextEdit can read selected text aloud.
 
  • Like
Reactions: stanleystf
chown33 said:
Thanks for the additional info.

Please post the exact command line you used in Terminal.

Did you try any apps other than Word? For example, I think TextEdit can read selected text aloud.
Click to expand...
Yes, I used TextEdit and it too sounds robotic and unnatural.

This is the command line I used:

say -f /Users/username/Desktop/textfile.txt -o output.mp3
 
bogdanw said:
SpeakLine – Text to Speech https://apps.apple.com/app/speakline-text-to-speech/id441968334
is free, uses the preinstalled voices, you can set the speed and export as aiff.
Click to expand...
I already tried this. Unfortunately, it sounds robotic and unnatural too.

Why is it that sounds unnatural, but this app sounds so natural?

apps.apple.com

‎Text to Speech-NaturalReader

‎With the Text Reader app turn your text into lifelike synthesized speech. Enable fluid, natural-sounding text to speech that matches the intonation and emotion of human voices. Text to Speech can export your tuned audio files effortlessly, thereby lets you create a high-quality audio file and...
apps.apple.com
 
I'm just guessing now, because I'm not running Sequoia here.

Please post the output of this command:
Code: 
say -v ? | grep en_
If this works as I hope, the output should be a list of English voice names. I want to see this list because I'd like to try a test running 'say' with the '-v' option to name a specific voice.


The info for the app you linked to says this:
Choose from over 300 of our high-quality, human-like voices across 129 languages and variants. Don't like what you're hearing? Change your settings anytime.
Click to expand...
It could be that the reason it sounds better is because it's using one of their high-quality voices.


If the choice is between paying for higher quality or being free with lower quality, you may have to choose the tradeoff you prefer.
 
  • Like
Reactions: stanleystf
chown33 said:
Code: 
say -v ? | grep en_
Click to expand...
The command returns “zsh: no matches found: ?”
Due to changes in zsh, it works with ? in quotes.
Code: 
say -v '?' | grep en_
Code: 
say -v '?' | grep en_US
Code: 
say -v '?' | grep en_GB

stanleystf said:
This is the command line I used:
say -f /Users/username/Desktop/textfile.txt -o output.mp3
Click to expand...

The default is AIFF and I’m not sure mp3 is even supported. For a list of supported formats:
Code: 
say --file-format '?'
It’s probably best to use it without the extension.
Code: 
say -v 'Ava (Premium)' -r 180 -f ~/Desktop/textfile.txt -o ~/Desktop/AvaPrem180
The default rate is 175, 175 words per minute.
 
  • Like
Reactions: stanleystf
stanleystf –

Au contraire to your earlier comment, I would say: converting arbitrary text to truly natural-sounding speech is an extremely complex task; it's taken decades for the technology to shift from passable robotic voices to the halfway-decent-but-still-synthetic voices that are (amazingly) included for free on the Mac nowadays. The diminishing returns associated with even more natural voices like NaturalReaders, which take all this server computing power even to use once, are still pretty close to the cutting edge.

So I think it seems reasonable to save up and pay for NaturalReader if you specifically like its voices that much, and would get that much value out of the difference between their product and another. It looks like NaturalReader is free to use on the Web for custom text on a personal basis, but they start asking you to pay for downloading the converted files.

Incidentally, I'm pretty sure Apple has been working on its own more natural text-to-speech technology for literature. We haven't heard much about that in a product, but I wouldn't be surprised if something like what you want actually does come included for Mac users this decade.

(I know you aren't keen on the system voices, but I'll add that it's pretty easy to use the Mac's built-in Shortcuts app to make a Shortcut that inspects selected text and saves an audio file to the desktop.)
 
Last edited:
  • Like
Reactions: stanleystf
stanleystf said:
I see. Can I use that Microsoft service to create Audio files myself?
Click to expand...
A quick search reveals edge-tts https://github.com/rany2/edge-tts

To install:

- install Apple Command Line Tools with
Code: 
xcode-select --install

update pip
Code: 
sudo -H pip3 install --upgrade pip

install edge-tts
Code: 
sudo pip3 install edge-tts

run edge-tts
Code: 
edge-tts --voice en-US-AvaNeural -f ~/Desktop/textfile.txt --write-media ~/Desktop/textfile.mp3
 
Last edited:
  • Like
Reactions: stanleystf
I've always used the Mac Automator app for creating audio from ebooks. Super simple, you just set up a workflow eg. get contents from an open text file + create audio file. Alex is my go-to voice until something better comes along. No, it's certainly not perfect and probably still too robotic for you, but for me had the most natural inflections for reading out of the Mac voices. You get used to its foibles after a while! (Had to redownload it as doesn't auto show on default Sequoia voice list). As @chown33 mentioned, it's either a case of using a free could-be-better voice or paying for something that is more natural. But I'm a cheapskate... 😆

Did a quick search to see if anyone has options to buy extra voices - came across CereProc voices (there's a few that don't sound too bad to be fair) but looks like they only work up to M1. It's a shame there are not more options like this that would give us access to higher quality and more realistic system voices considering Mac has default tools you can utilise for text to audio.
 
Last edited:
  • Like
Reactions: Starfia and stanleystf
chown33 said:
Please describe why you won't use the built-in features. If we know why it won't meet your requirements, then we can make better suggestions.

In particular, the 'say' command in Terminal has many options for voice, quality, rate, audio format, etc. See 'man say' for details.

Also remember that any Terminal command can be placed into a Shortcut or a scripted droplet, and have files of text dropped onto it for processing. That is, the "type a command" part can be automated and used repeatedly.
Click to expand...
I need to jump into terminal mode to create a script so that when I say “Siri,” it triggers Google Assistant instead. 🙋‍♂️
 
bogdanw said:
A quick search reveals edge-tts https://github.com/rany2/edge-tts

To install:

- install Apple Command Line Tools with
Code: 
xcode-select –install

update pip
Code: 
sudo -H pip3 install --upgrade pip

install edge-tts
Code: 
sudo pip3 install edge-tts

run edge-tts
Code: 
edge-tts --voice en-US-AvaNeural -f ~/Desktop/textfile.txt --write-media ~/Desktop/textfile.mp3
Click to expand...
I installed it. When I enter the last command in terminal, I get this error: command not found: edge-tts
 
Does this connect to Microsoft too?

apps.apple.com

‎Text to Speech: Voice Reader

‎With the Text to Speech app turn your text into lifelike synthesized speech. Enable fluid, natural-sounding text to speech that matches the intonation and emotion of human voices. Text to Speech can export your tuned audio files effortlessly, thereby lets you create a high-quality audio file...
apps.apple.com

It's a lot more natural sounding than the Mac's built-in voice.
 
stanleystf said:
I installed it. When I enter the last command in terminal, I get this error: command not found: edge-tts
Click to expand...
Close and re-open Terminal. Try again or with the full path
Code: 
/usr/local/bin/edge-tts --voice en-US-AvaNeural -f ~/Desktop/textfile.txt --write-media ~/Desktop/textfile.mp3

If you installed it without sudo, run:
Code: 
~/Library/Python/3.9/bin/edge-tts --voice en-US-AvaNeural -f ~/Desktop/textfile.txt --write-media ~/Desktop/textfile.mp3
 
  • Like
Reactions: stanleystf
bogdanw said:
Yes, and to revenuecat.com
Click to expand...
So the best TTS service out there is from Microsoft. I wonder why Microsoft Word on Mac doesn't use it! I've checked and Word on IOS actually uses it and sounds great! Problem is it can't covert the audio to Mp3.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back