Well they do have speaker identification. But I don't know how good it is. You may try them out. https://www.microsoft.com/cognitive-services/en-us/speaker-r... They also are having a private preview of CRIS which supports custom acoustic models https://www.cris.ai/