Remember when you were in infant school, sitting around in a circle, whispering a message from person to person until it reached the last kid in the chain and was completely different? As part of his "Digital Humour Theory" thesis, Pratt Institute graduate student Micahael J. Silber did just that with Siri's text-to-voice and Google's voicemail transcription services — 50 times in a row. Here's how he describes the four step process:
1. I recorded the audio of Siri reading a selection of text.
2. I placed a call to myself and played the Siri audio
recording into my Google Voice voicemail.
3. I instructed Siri to read the new Google Voice transcription,
including any errors and recorded a new audio clip.
4. I placed a call to myself and played the new Siri audio
recording back into my Google Voice voicemail.
As you can see in the short version of the video at the top, even after the first time through, Siri and Google Voice weren't doing such a great job. Granted, Silber gave the voice/text software a pretty big, wordy message to pass, but you'd think these programs from the two premier tech companies could do a better job than a room full of six year-olds.
Above is the abbreviated version, but if your life is devoid of entertainment for the next 30 minutes, feel free to subject yourself to the full 50 rounds below. It's not really necessary, though; it only takes a round or two to realise that what we have here is a serious failure to communicate.