AI audio player for websites
The team behind TYLR wanted to give publishers a way to offer articles in audio and reach readers in different languages. We helped turn that concept into a working MVP.
We developed the player and publisher dashboard, integrated speech synthesis and translation services, and built two ways to connect TYLR to existing websites: an embeddable script and a WordPress plugin.
Client quote
The product was delivered on time and as expected. Afterlogic.Works has successfully launched the product within the expected budget and timeframe. The team has good management and communication skills. Afterlogic.Works is responsive through email and virtual meetings.
About client
TYLR is developing a service that converts web content into speech and translates it for publishers and website owners. The founder approached us with a clear product concept and an understanding of the intended audience. We joined at the design stage and worked through to a functioning MVP.
About project
TYLR serves two groups of users. Listeners need a straightforward player: select a language, voice and playback speed, then start the article. Publishers need to install the service on their websites, customize its appearance and see whether their audience uses the audio option.
External services handle speech synthesis and translation. We connected text preparation, translation and audio generation to the player and publisher dashboard. Listeners access the audio version on the article page, while publishers manage the service through a separate interface.
Request
The client needed an MVP covering the complete user journey: connect a website, open an article, choose a language and listen. This involved four main tasks:
- Build a player with language, voice and playback speed controls.
- Configure the service to identify the main article text for narration.
- Reduce the wait before playback begins for longer articles.
- Create a dashboard with settings and analytics, and provide integration options for websites on different platforms, including WordPress.
Tools
TYLR builds on our experience integrating AI into working products. In other projects, we used a multimodal LLM to extract structured data from receipts and built an AI bot that turns news and legislative updates into drafts for human review. For TYLR, we connected speech synthesis and translation with an embeddable player and publisher dashboard.
What we did
Built a player for article pages
Readers select a language, voice and playback speed and listen without leaving the website. We connected the playback controls to text processing and external speech and translation services, bringing audio preparation and playback together in one product.
Isolated the article text
Article pages also contain navigation, advertising and related content blocks. Passing the entire page to a speech service would include text that does not belong to the article.
We configured the service to identify the main content and exclude unrelated blocks. This had to account for differences in website layouts, where the article and surrounding elements are arranged differently.
Processed longer articles in sections
To reduce the wait before playback, we organized text processing into sections. Playback begins while the remaining content is still being prepared.
Generated audio is saved, so listening to the same material again does not require another speech synthesis request.
Created a dashboard for settings and analytics
Publishers use the dashboard to customize the player’s appearance and settings, manage their accounts and view usage statistics. With settings and analytics in one interface, they can adjust the service after installation and track how their audience listens to published content.
Provided two integration options
The universal script allows TYLR to be embedded in websites built on different platforms. For WordPress, we developed a plugin to simplify installation and configuration.
The player runs within pages that already have their own styles and scripts. We isolated its interface to prevent website styles from affecting it and tested the WordPress plugin across different themes.
Tested the full listening experience
QA covered the complete journey from processing an article to playing the audio. We tested articles of different lengths and structures, translation and narration in different languages, and player behavior across websites.
We also tested the translation‑to‑speech workflow on real publications, assessing how clearly and naturally the translated text came across when spoken aloud.
Result
The TYLR team received a working MVP combining audio generation, translation into more than 30 languages, playback and service management.
Publishers can add audio to an existing website, choose an integration method and manage settings and analytics through the dashboard. They do not need to develop their own player or connect translation and speech synthesis services themselves.
Readers can choose whether to read or listen, in which language and at what speed. The startup now has a product it can demonstrate to publishers, integrate into their websites and develop further based on feedback.