Chinese internet firm Sohu's (Nasdaq: SOHU) online search subsidiary Sogou has released a real-time speech-to-text conversion app called "Sogou Tingxie" (literally "Sogou Dictation"), to fulfill long-form speech-to-text needs in situations such as meetings, interviews, and writing.
Sogou Dictation uses long-form speech-to-text technology exclusively developed for Sogou's speech recognition engine "Zhiyin" and integration of end-to-end deep neural network technology. Sogou Dictation supports usage scenarios such as writing articles, recording interviews, recording meetings, and daily note-taking. In response to different scenarios and needs, Sogou Dictation also supports two different modes, "dictation" and "transcription."
In dictation mode, text results are given in real-time as the voice recording is being made. In transcription mode, a recording can be input to the app, and will then be transcribed into text. Other features include editing while listening, wireless highlighting, syncing across multiple devices, and message sharing. In addition, when a person's speaking volume is lower than 30 decibels, Sogou Dictation can provide whisper recognition technology.
Editor's Note: For more information on this topic, please see "Sougou Launches Voice Recognition Engine Zhiyin," MD 8/03/16 issue.