Anyone who’s used voice recognition software will know that it’s more liable to misunderstand you if you are tired, or tipsy, or highly peeved at its inability to understand you. Also, lapsing into regional accents or lazy pronunciation – dropped h’s or t’s – can bring frustratingly odd interpretations of your words. The same is true of Siri’s voice recognition capabilities, and analysts from Piper Jaffray have compared Siri with Google’s voice search for accuracy.
Fortune magazine reports that even with the background noise of busy Minneapolis thoroughfares, Siri still managed a 62% accuracy rate. In a tranquil indoor environment, accuracy improved to 68%. Google on other hand turned in an excellent 86% indoors. Despite her 62% accuracy rate overall, Siri actually understood 83% of outdoor questions and 89% indoors, which means only around one in ten questions posed in quiet surroundings will be misinterpreted, so it’s fairly reliable. Under the covers, Siri collaborates with Wolfram Alpha’s VR software in only 14% of queries, while falling back on Google 60% of the time.