Autonomous learning strategies, supported by diverse media, have become essential tools for enhancing vocabulary acquisition. Among these, TED Talks stand out due to their accessibility, topical diversity, and authentic linguistic input. Although previous research has highlighted the educational value of TED Talks, limited attention has been given to their vocabulary characteristics across different topics. This study addresses this gap by examining the lexical profile, CEFR alignment, lexical level, and lexical density of TED Talks across 20 topics using a corpus-based approach. A dataset of 1,000 Talks (2,348,137 tokens) was analysed using computational tools and established linguistic frameworks. The findings reveal significant variation in vocabulary characteristics among topics, indicating their suitability for learners at different proficiency levels. Topics such as Relationships and Psychology cater to beginners; Education and Social Change are appropriate for intermediate learners; while advanced learners benefit from topics like Science and Government & Politics, which exhibit higher lexical density and complexity. The study highlights the importance of aligning TED Talk topics with learners' proficiency levels to optimise vocabulary acquisition. These findings provide practical guidance for educators and learners, supporting targeted language development through the tailored selection of TED Talks.