Last updated on January 16, 2020
Through the continuous advancement of technology, it is now easier to speak another language. Just recently, Google is introducing its Translatotron, the very first translation model that has the ability to convert speech directly from one language into another and at the same time, maintains the cadence and speaker’s voice. This device is also used to translate speech to text format and back to speech. It utilizes an end-to-end method and strategy that translates the voice of the speaker into another language. Google has high hopes that the development wills open up future developments that utilize the direct translation model.
Moreover, Google also states that Translatotron utilizes a sequence-to-sequence network model that performs task like processing input of the voice. It serves as a visual representation of the frequencies and creates a new visual representation in the desired language. The impact is considered to be faster with less likelihood. Another great thing about this tool is that it also works with an optional speaker component. This means that this device will work to keep the speaker’s voice. Though the device can synthesize the translated speech and may sound a little robotic, it can still keep or maintain some of the features of the voice of the speaker. Experts to ensure that it has the ability to deliver quality service and convenience to users have tested the quality and performance of Translatotron. The test was done by validating the translation quality through the measurement of BLEU score, computed with text transcribed by a speech recognition system. Despite the lag in the conventional cascade system, it is still evident that the feasibility if the end to end direct speech-to-speech translation has been clearly demonstrated. Moreover, compared to the audio clips that have been recorded, the direct speech to speech translation output processed by Translatotron has been highly appreciated by testers and experts.
The function and great features of Translatotron do not end there. Through the incorporation of a speaker encoder network, Translatotron has been able to retain the original vocal characteristics of the speaker in the speech that has been translated. This makes the translated speech sound more natural. This feature enhances the former Google research on the verification of the speaker and its adaptation. Hence, it is no doubt that using Translatotron, users can now have the freedom to retain their original voice.
If you are thinking that Translatotron is just basically the same as other translation tools and devices of Google. You are wrong. This new tool is far more different from traditional translator. This is because Translatotron bypasses the intermediate text representation steps so you need to expect that it is quite faster and more efficient. Google made this possible by utilizing a neural network that converts spectrogram of speech from one language into another language. Google also added a new approach by bring a lot of advantages that include faster inference speed, avoidance of compounding errors between translation and recognition and making it direct to retain the voice of the original speaker after translation.
Overall, we can fairly tell that Google has set the bar high again over its competitors. With its Translatotron, it will surely penetrate the market and support those professionals and people who need such tool. Translatotron is considered as one of the biggest achievement and creation of Google in the field of language translation. With its several benefits and advantages, it is no doubt that translatotron will be in high demand. Websites like https://seehearnow.org/muama-enence-translator-review/ are the best place online to go to for you to learn more about the best translators,