Even the most reliable ASR speech to text software can struggle to achieve accuracy rates over 80% and that’s under perfect speaking conditions with nice, clear audio. While ASR services allow you to transcribe even lengthy passages of text at great speed, there’s some bottlenecking to be expected in terms of accuracy. While certainly impressive, it’s important to remember that this process is not quite perfect… yet. This is then cross-referenced with the 40 human phonemes that make up human speech. How does ASR work?ĪSR services work by translating the vibrations we make in normal speech into digital language. Today, there are literally dozens of ASR services to meet all manner of transcription needs. Virtually every smartphone user makes regular use of ASR technology (whether they’re aware of it or not) through digital assistants like Siri and Cortana. Recent advances in digital technology have seen ASR software take off in a number of exciting ways. This early “digit recogniser” went by the name of Audrey and while she could only ascertain numbers, she represented a significant technological step forward for automated transcription.

In fact, the first example of Automated Speech Recognition Software (ASR) goes all the way back to 1952.

Speechnotes desktop