Can you play with AI voice recorders like this?

It is reported that Sogou held an online press conference to officially release two new AI voice recorders, S1 and E1. The S1 is priced at 2,698 yuan and the E1 is priced at 1,298 yuan. They were first sold on JD.com and Tmall, and two Forbidden City Palace co-branded models were also released simultaneously in cooperation with the Forbidden City Palace Culture.

Why does Sogou want to make an AI voice recorder?

When people mention Sogou, most of their impressions are "input method" and "search engine", but based on this, the company is planning a "language-centric AI strategy". Sogou founder and CEO Wang Xiaochuan once shared his observations on language AI at the 2019 China Business Leaders Annual Conference.

Wang Xiaochuan said that there will be two major development trends in voice artificial intelligence hardware products in the future. The first is from fixed equipment to mobility, portability and wearability; the second is IO (input and output) orientation, that is, through microphones, various sensors, GPS, magnetometers, etc., more data can be captured from the environment, and gradually from people adapting to machines to machines adapting to people.

Based on these two trends, Wang Xiaochuan said frankly, "Sogou's goal is to become a leader in the field of language artificial intelligence." "Language is the jewel in the crown of artificial intelligence." Wang Xiaochuan said, "We can even say that without language, we have no ability to be creative and reason. Today, everyone thinks that artificial intelligence can solve repetitive tasks because artificial intelligence does not have the creativity and reasoning ability, and because it does not fully understand language. This is the problem we need to understand."

Since 2012, Sogou has been developing core language AI capabilities around natural interaction and knowledge computing. In the past year, Sogou has made frequent moves in the field of voice recorders.

In March 2019, Sogou launched the AI ​​voice recorder C1. According to official sources, C1 is the first "new form" of AI voice recorder that integrates dual-microphone array, real-time transcription, cloud sharing and other functions. Since its launch, it has maintained the "No. 1 in total single product sales" on multiple mainstream e-commerce platforms. The upgraded version C1Pro launched later has also been well received.

In addition to the product itself, in August 2019, Sogou also joined forces with four industry companies, including Patriot, Newmine, Sony Voice Recorder, and Wancheng Group, to establish the AI ​​Innovation Alliance, and announced that it would open its dictation service to the entire industry, using its own AI technology to empower its partners.

Using AI technology to promote the transformation and upgrading of the traditional voice recorder industry is both cross-border and innovative, bringing new ways of playing to the voice recorder industry, which has not seen much new development for many years. In this process, Sogou not only developed the "AI voice recorder" category, but also successfully upgraded the voice recorder industry from three directions: products, technology, and industrial chain, by opening up dictation services and establishing an AI innovation alliance.

What else can you do with an AI voice recorder?

How to use the AI ​​voice recorder? Sogou believes that it can be an information acquisition tool that integrates voice, transcription, editing, storage and sharing. The S1 and E1 have achieved voice, transcription, editing, storage, sharing and translation functions.

But in any case, since it is a voice recorder, "recording" is the most basic function, and excellent sound pickup ability fundamentally determines the market performance and vitality of a voice recorder. Specifically for new products, the S1 has AI noise reduction function. It uses the pureVoiceAI noise reduction algorithm, which can filter more than 40,000 kinds of real noises, making the voice recording clearer. From the product introduction, it is not only equipped with 2 Harman directional microphones with a maximum pickup distance of 10 meters, but also equipped with 6 omnidirectional microphones, supporting 360° omnidirectional sound pickup. At the same time, based on Sogou's leading clairVoice8 microphone array algorithm, the S1 can bring users a 360° sound pickup experience in ultra-distant scenes without dead ends.

In addition to recording issues, more people are concerned about the shorthand arrangement after recording. Traditional voice recorders require repeated dictation after recording, which is a time-consuming and boring process. If you encounter memory or accent problems, it is even more troublesome. The transcription function of S1 and E1 solves these problems well. While supporting real-time transcription of recordings and transcription of recording files, it also has enhanced recognition capabilities, which can identify different speakers, applause, laughter, etc., and accurately distinguish and transcribe.

In addition, both AI voice recorders support synchronization with the Sogou input method vocabulary, and support recording and transcription in 10 languages ​​including Chinese, English, Japanese, Korean, and German, and 10 dialects including Sichuan, Guangdong, Tianjin, Shaanxi, and Guizhou. In addition, after a long period of training, the Sogou voice team has created language models in five professional fields, including finance and trade, medical care, IT technology, politics and law, and culture and sports. These language models are also applied to this new product to improve the recognition accuracy of industry-specific vocabulary.

Sogou said, "The transcription accuracy of both products is as high as 98%

Sogou believes that if super phonetic recognition is the basic item of AI voice recorders, and accurate transcription is the core item of AI voice recorders, then efficient organization is a plus point of AI voice recorders. In this regard, S1 and E1 use the "industry's first" NLP engine intelligent summary technology, which can organize paragraphs through intelligent semantics, automatically extract keywords to form tags, and intelligently extract paragraph summaries to facilitate user organization. You can also use voice to search for recording content in one sentence, and extract and summarize content based on the user's recording tags, applause, laughter and other nodes in the recording. Based on cloud storage technology, users can also automatically synchronize and manage recording data on voice recorders, mobile APPs, web pages, PC clients, etc., and realize convenient operations such as one-click export and scanning code sharing.

In addition to audio pickup, transcription, and organization, the extra "surprise" brought by S1 and E1 is undoubtedly their translation capabilities. They are the "first in the industry" recorders that support personal simultaneous interpretation, real-time Chinese-English translation, and WeChat mini-programs for multiple people to access and share translation content. S1 supports online translation in 63 languages ​​in 200 countries around the world, as well as offline translation in 9 commonly used languages ​​including Chinese, English, Japanese, Korean, French, and Russian. Its "industry-first" free dialogue translation function enables free communication in multiple scenarios.