Hot on the heels of the Wall Street Journal’s “The End of Typing: The Next Billion Mobile Users Will Rely on Video and Voice” article, Google has announced voice search in 8 additional Indian languages: Bengali,
Gujarati,
Kannada,
Malayalam,
Marathi,
Tamil,
Telugu,
Urdu.
This voice based dictation and search will be available on Gboard on Android, and in a search using the Google App, and more importantly, via Google’s Cloud speech API. They will be launched across other Google apps and products, including the Translate app. Where this development leads us is the introduction of Google Home in India, which is entirely voice based. Note that Gboard supports 22 Indian languages, so that’s still 13 languages to go, for enabling voice search/typing. Why is this important? Voice search is essentially a combination of two processes: the first is converting voice to text. This becomes complex as voices, tonality, phrasing and dialects change a language. As Google indicated in its blog post: “To incorporate the new language varieties, we worked with native speakers to collect speech samples, asking them to read common phrases. This process trained our machine learning models to understand the sounds and words of the new languages and to improve their accuracy when exposed to more examples over time. And voice input for each of these languages will get better over time, as more and more native speakers are making use of the product.” This part essentially covers the voice to text bit, but what improves results for Google is understanding context. It’s easier to transliterate words, but translation involves an almost human level understanding of the sequence of words. While Google hasn’t indicated this, but what has really sped up for Google now is
Bengali, Gujarati, Kannada, Malayalam, Marathi, Tamil, Telugu, and Urdu. This voice based dictation and search will be available on Gboard on Android, and in a search using the Google App, and more importantly, via Google’s Cloud speech API. They will be launched across other Google apps and products, including the Translate app. Where this development leads us is the introduction of Google Home in India, which is entirely voice based. Note that Gboard supports 22 Indian languages, so that’s still 13 languages to go, for enabling voice search/typing. Why is this important? Voice search is essentially a combination of two processes: the first is converting voice to text. This becomes complex as voices, tonality, phrasing and dialects change a language. As Google indicated in its blog post: “To incorporate the new language varieties, we worked with native speakers to collect speech samples, asking them to read common phrases. This process trained our machine learning models to understand the sounds and words of the new languages and to improve their accuracy when exposed to more examples over time. And voice input for each of these languages will get better over time, as more and more native speakers are making use of the product.” This part essentially covers the voice to text bit, but what improves results for Google is understanding context. It’s easier to transliterate words, but translation involves an almost human level understanding of the sequence of words. While Google hasn’t indicated this, but what has really sped up for Google now is translation. In April this year, Google had announced neural network translation in 9 Indian languages: Hindi, Bengali, Punjabi, Marathi, Gujarati, Tamil, Telugu, Malayalam and Kannada. At that event, Melvin Johnson, Engineer at Google Translate, had said that they’ve used neural machine translation to bridge the gap between phrase based (less efficient) and human (more efficient) translation. “Neural machine translation allows the translation of the entire sentence, instead of on a piecemeal basis.” Google builds language models, which allows translation from one language to the other, and with neural network processing, they’re able to translate without building one-to-one models. How does Google build these models? “Google has access to the entire web. We look for parallel documents on the web, which are trying to say the same thing in two different languages. For example, the BBC. We break it down and feed in examples of sentences in one language and in another, and it learns to do this mapping.”…“With neural systems, it takes 2-3 weeks to train per model, for hundreds for FPUs. For it to work really well, it needs hundreds of millions of examples for it to work really well. What does this combination of voice-to-text and translation enable? Contextual voice search across Google’s products and services, all of which integrate the Google Assistant: the Pixel phone, Google Home, along with supported devices like Chromecast. More on Google Assistant and its link with devices here. Give it a few more years, and all Android/linked devices and apps should be able to support usage in Indian languages, both search and commands.
Google Search and Gboard apps now support voice typing for 8 new Indian languages
Do you know you can talk to Google on your Android phone in Hindi? Yes, you can. Today Google is adding support for 8 more Indian languages to the voice typing and voice search on Android phones. The company says that starting today, Indian users will be able to also speak to the Google Search or the Gboard, the Google keyboard for Android phones, in Bengali, Gujarati, Kannada, Malayalam, Marathi, Tamil, Telugu, and Urdu. The apps earlier supported two languages — Hindi and English — in India. Google says that the support for new languages will help more Indian users to connect to the web using a language they are comfortable with. “With today’s update, users can now use voice input in their local language to search for what they are looking for without the need of typing on a tiny keyboard, making the user experience that much faster and easier,” said the company. Today’s update to the language abilities of Google Search and Gboard is a global update, says Google. In total support for 30 new languages has been added to the apps and of this 8 languages are from India. Daan van Esch, a technical program manager at Google, told India Today Tech that with its new language updates Google wants to make using Android phones and the web more comfortable to Indian users. Also Read: Inside Google Rail Wi-Fi in India and how it is connecting millions to the web “In India, there are over 300 million smartphone users and this number will grow in the coming years. We want to ensure that Google works for every India, whether he or she knows English or not,” says Daan. Google said that “in order to perform a voice-based search, users will need to set their language in the Voice settings menu in Google app”. If they want to use the local language to type their message using voice, the users will have to get Gboard keyboard app, which is available in Google Play Store. “Using voice to dictate a message is not just convenient, but it is also up to three times faster than typing. The new language support will also enhance voice typing on Gboard, helping users to respond to emails on the go and send texts within messaging apps,” says a Google spokesperson. Teaching machines to recognise Indian languages Google, along with several other big technology companies, is working to expand the accessibility to apps and services. As more and more web users come from regions where English, or even the Roman script, is not common, tech companies have tried to build support for regional languages in their apps and device. But it has not been easy. It is only now, when powerful machine learning tools are available, that companies like Google have seemingly cracked the code to built support for regional languages. Daan says that the support for 8 new Indian languages in the Gboard and Google Search is also powered by machine learning. This means, when you speak Tamil to your phone, it quickly records the data, sends it to the powerful Google computers that run neural networks and machine learning tools, understands it and gives an output to users. All of this happens within a second. “When a user speaks something to the phone, the audio is sent to ASR system. The ASR system has three components. First the sound is turned into a phoneme. The phoneme is then turned to a word. And the word then is given context to ensure it is accurate,” says Daan. The context part is very important. For example, without context, the phone may not differ between “you’re” and “your”. All of this work is done using machine learning, says Daan. He adds that the machine learning system is like a child. “In terms of difficulty, all languages are same. A child learns the native language naturally, however difficult or easy it is. A machine learning system is similar. It can learn any language. The key, however, is available content in that language. The more regional content is there on the web for the machine learning systems, the better it is,” says Daan. The Google technical manager adds that as more people use the Indian regional languages with their phones, the better it will get. Also Read: Our Wi-Fi at Indian railway stations is better than San Francisco, London Wi-Fi: Google Although in terms of difficulty, Daan says that machine learning systems find Indian languages as easy or difficult as any other language, there is one thing that is unique to Indian languages. Google notes that “while incorporating new (Indian) languages, (the company) worked with native speakers to collect speech samples, asking them to read common phrases”. This data that comes from the Indian users is always very noisy. “Voice samples that we collect from India always have these background noise that are unique to India. These samples have noise from Indian streets, and it is very identifiable. You can just listen to an Indian voice recording sample and because of unique background noise immediately identify that it is from an Indian city,” he says. Although background noise is a challenge, it is not insurmountable. Daan says that Google engineers ensure that the background noise is accounted for Indian users so that speech recognition works well and seamlessly. Google on Monday said that starting today, it will also be made available the new languages in the Cloud Speech API so that they can also be used in other Google apps and products, including the Translate app. “Google’s speech recognition now supports 119 language varieties in Gboard on Android, Voice Search and more,” the company said on Monday.
Google Voice Search adds support for eight more Indian languages
Google has expanded support voice search in more Indian languages. In addition to Hindi and English, Google will now support eight more Indian languages including Bengali, Gujarati, Kannada, Malayalam, Marathi, Tamil, Telugu, and Urdu. Starting today, speakers of these languages will be able to use their voice to dictate queries — both in Gboard on Android as well as in Search through the Google App. In order to perform a voice-based search, users will need to set their language in the Voice settings menu in Google app. The new language support will also enhance voice typing on Gboard, helping users to respond to emails on the go and send texts within messaging apps. To enable Voice Typing, users can install Gboard and choose their language from the Settings. You can tap the microphone to start speaking. Google says it has worked with native speakers to collect speech samples, asking them to read common phrases. This process not only helped train Google’s machine learning models to understand the sounds and words of the new languages, but also improved the accuracy when the system encountered more examples over time. Voice input for each of these languages is expected to get better over time, as more and more native speakers use the product. Google supports voice search for 119 languages globally and is adding support for 30 new languages, eight of which are from India. Google has been ramping support for Indian languages across its various products like Maps and Search. Voice search in eight new languages will be available in Google Search on iOS as well. These will soon be extended to other Google apps and products, including the Translate app. “Google Voice Search adds support for eight more Indian languages”, 5 out of 5 based on 18 ratings.