Language barriers are an important problem when it comes to healthcare services for minority groups, such as refugees or sign language users. Interpreters are not always available, especially in emergency settings. Alternatives, such as online machine translation, are unsatisfactory in terms of the languages they cover as well as their data confidentiality and translation reliability. To fill this gap, phraselators were developed in collaboration with medical staff. While these produce reliable translations, they remain unsophisticated systems: doctors have to search for questions using menus or keywords. To improve on this, the Geneva University Hospitals (HUG) have developed BabelDr, a speech-enabled phraselator. To see if speech improves functional suitability and usability, this study compares BabelDr against a phraseslator without speech (MediBabble). This is done by asking doctors to find precise information about a patient using the two tools in a crossover design. Results show that BabelDr allowed participants to collect most of the information in a faster and easier way than without speech.