How Apple’s Siri Learns a New Language

Print Email

When comparing personal assistants, Apple Inc. (NASDAQ: AAPL) stands well above the competition of Alphabet Inc. (NASDAQ: GOOGL) and Inc. (NASDAQ: AMZN) in one key area: language. Although most of these assistants are similar in many ways, language is a dividing factor, yielding a larger target audience.

Apple’s Siri can support 21 languages across 36 country dialects. Google’s Assistant is only capable of understanding four languages, while Alexa from Amazon only knows two. Even Cortana from Microsoft Corp. (NASDAQ: MSFT) only supports eight languages.

With Apple’s next update, iOS 10.3, the company is introducing another language, Shanghainese, further expanding its stranglehold on this field.

Alex Acero, head of the speech team at Apple, illustrated the process of how Siri learns a new language in a recent interview with Reuters.

Siri begins learning a new language after Apple brings in humans to read passages in a range of accents and dialects, which are then transcribed by hand so the computer has an exact representation of the spoken text to learn from. Apple also captures a range of sounds in a variety of voices. From there, a language model is built that tries to predict word sequences.

The next step in the process involves Apple’s “dictation mode,” a text-to-speech translator in the new language. When customers use dictation mode, Apple captures a small percentage of the audio recordings and makes them anonymous. The recordings, complete with background noise and mumbled words, are transcribed by humans, a process that helps cut the speech recognition error rate in half.

Accordingly, when enough data has been gathered and a voice actor has been recorded to play Siri in a new language, it is released with answers to what are expected to be the most common questions. Once released, Siri learns more about what real-world users ask and is updated every two weeks with more tweaks.

Shares of Apple were last seen at $138.60 on Thursday, with a consensus analyst price target of $142.46 and a 52-week trading range of $89.47 to $140.28.

Alphabet shares were trading at $855.90. The stock has a 52-week range of $672.66 to $867.00 and a consensus price target of $988.95.

Amazon was trading at $852.38, in a 52-week range of $538.58 to $860.86. The consensus analyst target is $942.32.

Microsoft shares were last seen at $64.91. The consensus price target is $69.48, and the 52-week range is $48.04 to $65.91.