At its Made By Google 2023 event today, Google has unveiled the Pixel 8 and Pixel 8 Pro. As has been the trend nowadays from most of the big tech companies, most of the event was about showing off its AI capabilities and new technologies that the new Pixels bring.
At the heart of the Pixel 8 and the 8 Pro is the new Tensor G3 system on a chip (SOC), and the new powerful TPU or Tensor Processing Unit which makes all the AI processing possible. According to Google, the new Tensor G3 is capable of doing twice the machine learning (ML) processing as compared to the original Tensor on the Pixel 6 device and much of the credit for this heavy lifting is due to the TPU.
 |
 |
If you are wondering what this means for the overall performance of the Pixel 8 and the Pixel 8 Pro, the end result is more advanced AI capabilities and Google demoed some of these features today.
First up, Google Assistant has got smarter. Now, the Assistant will be able to understand your speech and what you want from it even if there are pauses when you say it.
Google says "Now you’ll also be able to talk to Google Assistant more effortlessly. You can speak at your own pace — it even understands pauses and “um”s. We combined Google’s state-of-the-art speech recognition with server-quality national language understanding." Besides being a better assistor, the Tensor G3 also brings better translations wherein it can do real-time translations due to improved speech recognition.
There are also improvements related to audio processing and noise cancellation. By combining the new Super Wide Band, which doubles the bandwidth, with the Clear Calling feature, the voice clarity via Bluetooth is improved greatly while noise is eliminated. The two before and after images below show the difference in audio bandwidth:
Another feature related to noise cancellation is the Audio Magic Eraser which helps to eliminate unwanted background sounds like noise from wind, dog barks, among others. In the example (below), Google demonstrated the feature by eliminating a dog's barking from the background so that the captured video only plays the playful voices of the baby.
 |
 |
A cool new feature for taking photos is "Best Take" which according to Google "uses a diverse set of advanced machine learning models on-device to combine multiple images and help everyone look their best". Essentially, what it does is allow the user to choose another expression from a multitude of options if they aren't too happy with the originally captured frame.
Spam calls can also be smartly eliminated with the Call Screen feature. The AI will have the ability to process and understand the intentions of the caller. If the AI detects that the caller is a potential spammer, the voice assistant cuts the call. In case, the call is a legit one, the assistant proceeds to the alert the user.
 |
 |
Google claims that Call Screen feature can lead to "50% fewer spam calls on average" which means there might be room for some real calls being detected as spam.
2 Comments - Add comment