Simon is about 90%-95% accurate and many users have written to us claiming it is near perfection.
Artificial intelligence technology has made huge advances but it is not perfect and not right for every situation.
The greatest determinant of accuracy is the quality of the recording. If your audio has a lot of background noise, your transcript's accuracy will suffer.
If there is little/no background noise, you use lapel mics, and one person speaks at a time: you will get better results.
See here for our guide to get accurate transcripts.
We encourage new users to use the free credit to try their file with our speech recognition system and make sure they are happy with the results first.
See the difference clear audio makes with the videos below. Turn on captions to see the unedited Simon Says subtitles for both versions (note also the gaps in the transcript due to background noise in the noisy sample).
Related Help Articles:
Are there file size or length limits when uploading to the website?
Can I purchase in advance and prepay for transcription / translation credit?
What export options do you support?
What is the optimal way to record audio for transcription?
What methods can I import files to your website?
Which formats/codecs do you support?