In Part 1 of our Dergilik journey, I discussed expanding content options for our users. We recognized that while Dergilik offered a rich library of articles across diverse topics like travel, food, and automotive trends, users often have moments when reading isn't ideal. Consider users doing chores, commuting, or engaging in activities that occupy their hands and eyes. Our goal was to provide an alternative consumption method: an immersive listening experience. 🎧 We set out to bring Text-to-Speech (TTS) capability to Dergilik articles, transforming them into audio experiences.

Understanding the Linguistic and Technical Challenges of Turkish TTS

When we decided to implement TTS, we knew that simply selecting an off-the-shelf engine might not suffice for delivering the quality our users deserved. Turkish, with its unique linguistic characteristics, presents specific challenges for TTS systems. From a technical standpoint, Turkish grammar, including vowel harmony, agglutinative morphology (word formation), and the accurate pronunciation of Turkish-specific characters, demands a nuanced and tailored approach to TTS engine design. Generic TTS engines often fall short in capturing these subtleties, resulting in robotic or unnatural-sounding speech.

Strategic Evaluation and In-House AI Advantage

Therefore, our evaluation process for TTS engines went beyond basic functionality. We rigorously tested engines from Google, Microsoft, and Nuance, specifically assessing their performance with Turkish content. While these engines offered a baseline level of Turkish support, our internal AI team, with their deep understanding of Turkish linguistics and machine learning, demonstrated the potential to create a truly superior, purpose-built Turkish TTS solution for Dergilik. Choosing our AI team's in-house engine, despite it being pre-production, was a strategic product decision driven by the pursuit of exceptional quality and a differentiated user experience.

Mobile App Development: Iterative Design and User-Centric Features

We prioritized mobile app integration, as visual examples best illustrate the user-facing improvements. Leveraging the expertise of Dergilik's mobile developers, we also collaborated with the fizy (music streaming app) mobile team to accelerate development and tap into their streaming knowledge. Turkcell’s vast talent pool proved invaluable, providing readily accessible experts in both iOS and Android ecosystems.

Let's examine the evolution of our mobile player. (See images below - Keep these if you have them). The image on the left represents our initial version 🙈, and the image on the right showcases the current, enhanced player 🚀.

The first iteration, while functional, lacked key user-centric features. Specifically:

• No Playback Speed Control: Users were locked into a single playback speed.
• Limited Sharing Options: Content sharing was unavailable.
• No Playlist Functionality: Playback ceased after a single article, disrupting the listening flow.
• No Seek Functionality: Users could not easily navigate within the audio.

The evolved player addressed these limitations, resulting in a significantly more user-friendly and feature-rich experience. Our development philosophy for the mobile app was rapid iteration. Instead of striving for perfection in the initial launch, we prioritized releasing a basic functional version quickly and continuously enhancing it based on user feedback and A/B testing. This iterative approach allowed us to deliver value to users sooner and adapt to their evolving needs.

CMS & AI Integration: Content Workflow and Quality Assurance)

Simultaneously with mobile app development, we collaborated closely with our Content Management System (CMS) and AI teams to establish a seamless content pipeline. We understood the content team’s workflow:

• Daily Content Creation: The team produces approximately 10+ new articles across diverse topics daily.