Figure S3. Speech recognition results of direct REAL audios from masks (supplementary audios) through Google speech-to-text platform.