Automatic speech recognition systems, sometimes referred to as speech-to-text technologies, have revolutionized clinical documentation with real-time transcription of spoken language into structured text. The current status of ASR applications in medical documentation is discussed herein, with the primary focus being radiology. Results demonstrate that ASR systems significantly reduce times for documentation turnaround and improve workflow efficiencies while decreasing costs related to transcription. However, some drawbacks still appear in speaker variability, complex medical terminologies, and accuracy for non-native speakers. Works emphasize indeed remarkable achievements of research in the improvement of ASR performance by including structured templates and the use of machine learning and natural language processing approaches; this technology is not at a human transcription level considering precision and usability. Economic factors, implementation issues, and user acceptance make the difference in the implementation of ASR systems. This review has underlined the transformative potential of ASR technologies while calling for further research in addressing the existing limitations and optimization of their integration into clinical workflows.