Multilingual automatic speech recognition (ASR) is crucial for global communication, with solutions like OpenAI's Whisper and Gladia's innovative approaches advancing the field. Historically, ASR systems relied on acoustic, lexicon, and language models but faced challenges with accents, dialects, and language nuances. The evolution from statistical models to deep neural networks and transformers has improved language detection and transcription accuracy. Whisper, for example, uses a transformer architecture for multilingual capabilities and has been trained on extensive audio data to support over 99 languages. Despite advancements, challenges remain, such as handling low-resource languages, accents, and code-switching. Gladia addresses these issues by utilizing a hybrid system combining machine learning and rule-based approaches to enhance language detection and manage accents. Their API supports over 100 languages, providing transcription, diarization, and translation, aiming to make multilingual ASR more accurate and accessible for various applications, including virtual meetings and call centers.