MacSpeech this week at Macworld Expo San Francisco launched its latest speech-to-text software, Scribe. Instead of the real-time recording and translation in its Dictate packages, Scribe handles recorded audio files.
The $149 software understands 13 different English dialects, the company said, and users can train the software to improve accuracy. MacSpeech Scribe accepts audio files in .wav, .aif, .aiff, .m4v, mp4, or .m4a formats.
Watching a demonstration on the show floor, the software first required users to verify the accuracy of transcription with several minutes of the speaker's audio. The program highlighted a sentence or phrase of translated text and then played the audio for verification. It appeared to be accurate.
However, the software is designed to transcribe only a single voice at a time. So, it's not good for a meeting where a number of voices speak one after another. However, it would work for a speech. The company said that a single MacSpeech Scribe license will let users create up to 6 individual voice profiles.
Certainly, this will expand the capabilities for persons who record notes remotely on the iPhone and then want to turn that audio into text. The company advised that background noise and the quality of the recording will affect the accuracy of the translation.
Registered owners of MacSpeech Dictate 1.5 can purchase Scribe for $99.