Intelligent voice technology has been applied in the broadcasting industry since the end of last century. With the popularization of deep learning algorithms, it has gained strong technical support. At present, there are four applications of intelligent voice in the broadcasting industry, namely, broadcasting content collection, broadcasting content generation, broadcasting content dissemination, and broadcasting content management. This article briefly summarizes this to clarify the progress of the next step.

The application level of intelligent speech technology in the broadcasting industry is based on the development level of intelligent speech technology. After 2006, with the popularization of deep learning algorithms in the field of intelligent speech research, the technical level of intelligent speech has only developed rapidly. The application of intelligent voice in the broadcasting industry has also spread rapidly, and at the same time it has begun to advance in depth.

1. Application of Intelligent Voice in Broadcast Content Collection

The editing of traditional broadcast content mainly relies on reporters to record on the spot with voice recorders, computers, and paper and pen. However, it is often time-consuming and labor intensive to listen to the recordings repeatedly when collating interview materials. The emergence of intelligent voice provides a solution to this problem.

In 2015, the former Central People's Broadcasting Station and the Chinese Academy of Sciences signed an agreement to jointly build a China Broadcasting Cloud Platform. One of the important contents is the establishment of an all-media acquisition and editing system for domestic radio stations to realize the evolution of broadcast news acquisition and editing in the era of artificial intelligence. In the all-media acquisition and editing system, reporters can use the mobile phone acquisition and editing client to convert interview recordings into text in real time. The recognition accuracy is above 95%, and semantic modification can be carried out. Basically, it has reached the standard of instant acquisition and publishing, which greatly improves Improve the efficiency of news editing. This system supports 5 minority languages ​​such as Mongolian, Tibetan, Uyghur, Kazakh, and Korean, as well as Chinese, English, Russian, French, Japanese, Korean and other languages. It can be used to record the above through the intelligent voice engine. Real-time translation of the language into Chinese breaks through the limitations of the reporter's personal ability and makes the "impossible interview" possible.

However, this omni-media acquisition and editing system still has limitations. For example, it needs to rely on mobile phones for recording, and the effect is not ideal in noisy interview environments or on-site long-distance interviews; it also needs to rely on mobile phone network signals and encounters complex network scenarios (such as In large-scale conference scenes) or in scenarios where there is no network (such as remote mountainous areas), the voice conversion capability is also greatly reduced, and voice translation is basically unavailable.

There are actually many similar professional smart voice devices, such as the Livescribe smart pen produced by Sweden's Anoto. It allows reporters to record key messages with a smart pen while recording an interview, and synchronize the recording and text to the cloud. At the end of the interview Later, when you click on the relevant text in the mobile application, the live recording at that time will be played, which will be beneficial to the audio data query and the production of radio programs. There is also the Sogou Smart Voice Recorder C1 launched by Beijing Sogou Company. Based on the basic voice-to-text function, it also completes functions such as automatic segmentation, tone particle filtering, keyword optimization, online editing, and key marking.

As more and more radio stations begin to equip reporters with these professional-grade intelligent voice interview equipment with compact size and complete functions, intelligent voice technology is playing an increasingly important role in the collection of broadcast content.

2. The application of intelligent voice in broadcast content generation

Although the earliest CRT oscilloscope appeared in 1897, oscilloscope technology was widely used in broadcasting content production after the 1960s, which made the content collected by broadcasting become "visually visible." Although this is a huge leap in the production of broadcast content, it also has the original flaw of "visible but incomprehensible". Program producers often need to spend a lot of time repeatedly listening to confirm the specific content, so the efficiency cannot be maintained. To solve this problem, broadcasting must be changed from "visually visible" to "visually understandable", and intelligent voice can meet this point.

In July 2017, Jiangsu Radio and Television General Station launched a content compilation system internally for the post-production of radio programs. This system can convert broadcast audio files into text through intelligent voice, and one-to-one correspondence between audio content and text content, and a one-to-one correspondence between the speaker's voice and a specific color. This system can achieve the following functions: one is to support audio addressing through text, which allows editors to quickly find and edit the required audio clips; the other is to support preset sensitive words and specific broadcast content containing sensitive words Color marking, which will help editors conduct program review; third, support the production of video content subtitles, and provide convenience for the integration and dissemination of sound, text, and video by radio stations on new media platforms. Observing the operating logic of this system, we will find that intelligent voice is the core of its many functions. It transforms audio waveform editing into document editing and even graphics editing through the content conversion of sound and text, which fundamentally breaks media editing. The boundary of broadcasting content has changed the way of post-production of traditional broadcasting content, and has greatly improved the efficiency of broadcasting content production. It should be said that this system of Jiangsu Radio and Television General Station is a successful application of intelligent voice technology in the background compilation of broadcast content production. However, this system has not yet been promoted. No matter at home or abroad, similar attempts have been made. Relatively rare.

At the same time, another application of intelligent voice technology in the production of broadcast content has appeared more widely in major broadcasting stations. This is the participation of anthropomorphic intelligent voice roles in the front-end broadcast of the program. The most representative is the "XiaoIce" intelligent voice technology developed by the Microsoft (Asia) Internet Engineering Institute. From 2017, Beijing Radio and Television Youth Broadcasting FM98.2, Hunan Radio and Television Music Voice FM89.3, and China Central Radio and Television Station Guoguang HitFM FM88.7 three broadcasting frequencies started. As of August 1, 2019, "XiaoIce" has hosted 63 radio and television programs in China. In these programs, "XiaoIce" either partnered with a live host, or hosted the whole process by himself, and his host level was amazing. For the production of broadcast content, the benefits brought by the addition of "XiaoIce" are obvious: First, it can enable radio stations to truly achieve all-weather, uninterrupted high-quality broadcasts, and fully guarantee the production and broadcast of radio programs. Continuity. The second is to ensure that accurate information is provided. Because "XiaoIce" has a very rich knowledge reserve, and there will be no oral broadcast errors, which ensures the stability of the production and broadcast of radio programs. Third, by understanding the preferences of the audience, "XiaoIce" can change from person to person and provide users with accurate content services in real time, which improves the pertinence of radio program production and broadcasting. Fourth, it is able to enrich the style of broadcast programs. In addition to being an anchor, "Xiaobing" can also be a news commentator, singer, poet, etc. It can sing a few words or comment on current events in the radio program at the invitation of the audience, or write a first for a caller. Hidden head poems, these functions are difficult for an ordinary live host to have at the same time. Because of "XiaoIce"'s outstanding performance, more and more domestic radio stations are sending out invitations to it.

3. The application of intelligent voice in broadcasting content dissemination

In June 2019, National Public Radio (NPR) and Edison Research released a "Intelligent Voice Report". The report showed that 21% of American adults over 18 years old Of people own a smart speaker, the number of owners reached 53 million, and this number was 14 million in 2018, and its growth rate is as high as 378.6%. Among those surveyed who own smart speakers, 42% use smart speakers to get news, 37% listen to AM/FM broadcasts, and 55% think they have listened more after owning smart speakers 74% and 66% of the audio content said they would use smart speakers when doing housework or cooking.

Although this report characterizes the current situation of the use of smart speakers in the United States, it is representative and at least illustrates several important roles of smart voice in the dissemination of broadcast content: First, it expands users' radio listening channels. This makes it possible for broadcasting to have more channels of communication in addition to traditional broadcasting radios and smart phones. In China, the number of users of smart speakers in the Chinese market in 2019 has reached 23.7 million units, an increase of nearly 8 times compared with 2.57 million units in 2018.

In the United Kingdom, there were already 9.5 million smart speakers in 2018, an increase of 98.6% over 2017, and this number will increase by 31.6% in 2019, reaching 12.6 million. The second is to increase the user's radio listening frequency. People are willing to turn on smart speakers with simple wake-up words, and use convenient voice interaction to choose their favorite broadcast frequency, and let it play accompanying. The third is to make broadcasting re-enter the family. In the living room, smart speakers are becoming more and more essential appliances like TVs. They bring a family together to listen to the radio, just like a family listening to Roosevelt’s "fireside conversation" on the radio in the 1930s in the United States. .

Intelligent voice technology has opened up new markets for broadcasting content, and has attracted more and more broadcasting stations to connect their frequencies with smart speakers in order to reach more users. As early as 2014, National Public Radio cooperated with Amazon Echo smart speakers to become its default news provider, and then stationed in Google Home smart speakers and Apple Homepod smart speakers to provide them with broadcast audio content; NBC Universal Group, USA Consumer News and Business Channel (CNBC, Consumer News and Business Channel) began to provide news services for Amazon Echo smart speakers in December 2016; the British Broadcasting Corporation (BBC) also announced in December 2017 that Amazon Echo Smart Speaker users can listen to the contents of the BBC’s 56 radio frequencies on this platform. At the same time, popular programs such as "Ache's Family" and "Women's Time" have been moved to it. At the same time, it is worth noting that, due to the presence of these intelligent voice platforms, the content produced by broadcasters can be found by users in a voice-controlled manner without barriers.

The above-mentioned applications are very important, although this connection is essentially just an extension of the channel for broadcasting content. In fact, on a new sound consumption platform such as smart voice, the industry is more concerned about exploring a form of sound transmission and consumption that is completely different from radio stations and Internet radio stations. This is the new voice that smart voice brings to the dissemination of broadcast content. think.