LogoReadvox
Back to Blog

Readvox Usage: respectful use of text to speech

Resources are vast but they are not unlimited. We all need to share this amazing text-to-speech functionality responsibly.

TLDR:

sample

If you see a blue bar:

You are ok. Thank you for not overusing resources and being a good citizen. This makes it possible for others to enjoy free text-to-speech narration.

If you see an orange bar:

You are at a warning level of extension usage for the day. This indicates significant strain on the server due to your above-average usage.

If you see a red bar:

Please take a break. You are critically overusing the resource, causing other users to experience pauses in the audio stream and putting critical strain on the server.

How Readvox Works

I’ve put a lot of skills, knowledge, and effort into optimizing speech production. It’s a unique combination of trained AI models and processing units that makes it possible to serve many text-to-speech generations to users at no cost. But that’s not the full story.

AI models don’t work on their own; they run on servers, infrastructure, domains, and many other behind-the-scenes parts. There’s also a lot of effort put into the design and development of the extension itself, but that’s a story for another blog post.

When you press play in the Readvox extension, it analyzes what content makes sense to read. Usually, pages have special areas tagged as Article or Main. If you ask to read a specific part of the text by selecting it, Readvox takes directly what was given to it.

The next phase involves Readvox chunking the content into smaller parts for privacy and effective processing. Then the extension passes these chunks to processing units located in different places. These units quickly convert text to speech and return the generated sound files to the extension. Finally, the audio files are played in the right order. This is an oversimplification, but it provides a general understanding of the inner mechanics.

How About My Data Privacy

I take privacy very seriously and would never want someone to be aware of what content I’m listening to. The same approach applies to Readvox text-to-speech generation. The text is encrypted and split into chunks, delivered through different channels. Different parts of the same sentence are delivered to different processing units, where they are passed to AI models solely capable of converting text to speech. Once conversion is done and the audio file is sent to you, the text is deleted from the AI memory, and the audio file is deleted after upload. This ensures that even a text fragment exists in the processing unit for a fraction of a second and then is irreversibly deleted.

Is It Going to Be Paid Sometime in the Future?

I’ll always prioritize the initial idea. When I felt lost trying dozens of different extensions from the Chrome Store, each with robotic voices or limited functionality, all I wanted was to read a blog post with breakfast. I’ll always keep Readvox free for moderate use.

For cases where some users play hours and hours of content, putting a critical strain on the system and negatively impacting others, I’ll consider measures and guidance to help everyone be a good citizen. The first measure is the “Usage” tab, coming in version 2.4.0

Current load

Even now, while writing this post, I see that the strain on computing units is way above moderate use.

Here is an example of the load on one of the processing units:

Processing unit load example

There are already peak loads, meaning the processing of text-to-speech for some users is delayed. This means some users overuse the extension far beyond moderate use. But how do you know if you are overusing it and causing bad experiences for others?

How to use respectfully? The Usage tab!

The purpose of this tab is to show how your usage impacts the shared infrastructure and affects others.

When you install the extension, it will show 0 characters for each of the last 7 days.

Initial usage tab

Once you start using it, the last seven days will start filling with vertical bars representing the number of characters converted via text-to-speech each day.

Blue level of load, all good

While you see blue bars here, it’s all good. You are being a good citizen and using it responsibly.

Some days, you may see the play button turn orange. If you click on the speaker avatar, you’ll see an orange dot in the Usage tab. Click on it, and you’ll see that you’ve reached the Overuse state for the day. This is a warning signal that you are getting closer to a critical state of free service utilization. You can still proceed with listening, but be aware that you are approaching the Critical state. This usage is still considered responsible.

Overuse level

If you continue to increase usage after seeing the orange “Overuse” state, the play button will turn red. In the Usage tab, the usage bar for the day will show a red part on top, signaling that you’ve crossed the Overload capacity and are currently negatively impacting other users.

Critical level

What If I Still Want to Continue Listening?

The Readvox community is friendly and always tries to be helpful and responsible. Many users rely on Readvox due to visual impairments. Students need to finish a study paper before an exam. If it’s a critical situation, please continue. You are your own judge, and all I ask is for you to use it responsibly.

Why It All Matters

I’m trying to balance this system of providing free service while maintaining the high quality. I ask for your help in keeping this balance. If this model is stable, I can continue developing Readvox in my free time. This means new voices, new languages, and new features will keep coming into the app. Moderate usage will remain available for everyone who once felt frustrated by the lack of good text-to-speech extensions and then found Readvox.

author
Anton
Founder

Drop me a note

FAQ

Back to Blog