Listen to your own beautiful voice to navigate, this is the new function of Baidu Maps

A wise artist once said: I have been in many cars, including some very bad ones. I have experienced this deeply. The driver took the wrong road, played disco music in the car, smoked cigarettes in the car, etc. What's worse, not only is it equipped with the above slots, but the mobile phone navigation voice also uses a very unpleasant voice. Whenever this happens, I want to help him replace it with a pleasant voice, such as my own.

I believe that many drivers’ navigation voice packages are either Zhiling or Guo Degang. Although they sound nice and interesting, they will inevitably become boring over time. The new voice customization function of Baidu Maps not only solves this problem, but also increases the author’s stickiness to this app.

The traditional speech synthesis process (such as AutoNavi's World of Warcraft voice package) usually requires recording more than 1,000 sentences, and the synthesis also takes a very long time (the production cycle of AutoNavi's World of Warcraft voice package is as long as 3 months). However, Baidu Map's voice customization can be completed in 20 sentences, and the synthesis time is also very short.

When you upgrade Baidu Maps to the latest version, you can see a "Voice Customization" icon in the upper left corner. After clicking it, you can choose the voices of stars such as Hua Chenyu, Qin Lan, Meng Dan, etc. In the recording page, you can choose female voice, male voice, female child voice or male child voice, as well as 4 types of recording texts according to the attributes of different users.

However, after starting to record, you need to find a quiet environment. There are a total of 20 recorded texts. It takes about 5 minutes to complete the process. The entire recognition process is very smooth.

After recording, the voice will enter the synthesis stage, which takes a total of 15-20 minutes and is done entirely in the cloud, so even if you exit the app it will not have any impact . After completion, just click to download. The author's voice package capacity is 10MB, which takes up almost no storage space.

However, due to factors such as the recording environment, sound characteristics, and the number of people recording, there may be a certain delay in the production of voice packages for some users.

After the download is complete, you can hear your own AI voice , and at this point, all you need to do is start navigating. When I first heard the navigation voice, I was really shocked. Although it was not realistic, the similarity was quite OK, and there was a sense of familiarity that came over me, forming an illusion that there was another me next to me.

And you can use your own voice not only when navigating, but even when interacting with Xiaodu Assistant, such as asking about weather conditions, nearby food, navigation to the Eiffel Tower, etc., it also uses synthesized voice for feedback instead of the default voice. This adds a sense of intimacy and makes me more willing to use this feature, or even the entire map app.

But listening to it yourself and to others is a completely different experience, so I also let people around me listen to the voice synthesized by Baidu Maps. Everyone's first impression was the same: This is indeed my voice. In terms of similarity, everyone said it was around 50%-70%.

However, I think this function is not for "yourself" use, because listening to your own voice for a long time is also quite boring. Personally, I think it is more for you to record the voices of important people around you. If you are a husband, you can listen to your wife's voice on the long commute to work in traffic jams; if you are a parent, you can listen to your children's voices. In addition, there are many more interesting usage scenarios for users to explore.

Furthermore, if your voice is nice enough, you can even publish the synthesized voice to the Baidu Map platform for others to download and use, so that you can better interact with other users. I browsed around and found that many people would imitate celebrities or Crayon Shin-chan, which is really interesting. After making a voice package, you can also share it with friends around you who are also using Baidu Maps. You don’t need to use other people’s phones to do it. You can quietly change it for your relatives and friends, so that it can also be a small surprise.

In the past, listening to your own voice while navigating seemed like a distant thing. As the world's first map app that supports voice customization, Baidu Maps can make it happen. After using it for two days, I still like this feature of Baidu Maps very much. The whole process takes less than 30 minutes , while traditional speech synthesis technology usually requires recording up to 1,000 sentences, and the synthesis time is also very long. The reason why Baidu's voice customization is so fast is that Baidu uses the original style transfer technology Meitron model, which has good performance in many aspects such as timbre, emotion and rhythm, and is currently the industry's leading speech synthesis technology.

This further positions Baidu Maps as an AI map, and more importantly, this feature does not require you to record in a professional studio, a quiet environment is enough. No matter what product it is, ease of use is very important, and throughout the whole process, I did not feel any threshold, from recording to use, it was done in one go, very simple.