BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

616 

Analysis of the Artistic Elements of Broadcast Hosting Based 
on Media Speech Corpus 

Zexuan Huang 
Communication University of China, Beijing, China 

Abstract. Whether it is speech synthesis research or speech recognition research, it depends on 
the construction of excellent speech corpus at the back end to some extent. The phonetic corpus 
includes three forms: text database, phonetic database and acoustic parameter database. Based on 
phonetic facts, it serves the basic research of language and the processing of language 
modernization technology through the functions of natural phonetic collection, phonetic annotation, 
retrieval and statistics. The purpose of the construction of the media corpus is to serve the research 
of the audio language of the media, especially the prosody research of the host language. 
Broadcasting and hosting is an important part of radio and television, and the diversity and richness 
of modern radio and television programs also put forward new requirements for broadcasting and 
hosting. Broadcasting determines the basic style, purpose and orientation of radio and television 
news programs. In the media industry, the status of broadcasting and hosting art is gradually 
highlighted. Under the background of media convergence, broadcasting and hosting art needs to be 
developed in more advanced technical fields, and the media voice corpus is an important product 
created by the development of science and technology.  

Keywords: Media; Phonetic corpus; Broadcasting and hosting art; Data processing. 

1. Introduction 
With the development of corpus linguistics, it has been widely recognized that it plays an important 

role in the field of language research and natural language processing, and various forms of corpus 
construction have also made great achievements. The use of corpus in various fields has become very 
common. The media language corpus under construction is a comprehensive corpus system featuring 
audio media and radio and television corpora, which includes two closely related sub-corpora, the 
phonetic corpus and the text corpus. To build a framework of speech database, we should first 
consider its purpose. The speech database used for speech recognition is different from the speech 
database used for speech synthesis. Speech database for speech recognition is to provide training data 
for acoustic layer. It not only involves linguistics, but also needs to comprehensively reflect the 
acoustic characteristics of speech from the perspective of phonetics and acoustics, so as to provide 
complete data for establishing speech models. With the continuous development of science and 
technology, the mode of cultural transmission has also changed, from the original simple printing 
process to the current cultural feast of visual and auditory coexistence. Under the guidance of 
phonetics and linguistics, the design of a scientific and reasonable continuous speech database is of 
great support to the speech recognition and synthesis system. 

We know that in the past, among all kinds of media, TV was the most important media to spread 
art for a long time, but today, with the continuous development of the times, the network has 
penetrated into thousands of households and is closer to people's lives. Moreover, the art of 
broadcasting and hosting needs to show its unique charm through a wide space. Corpus is an effective 
and indispensable research means and plays an irreplaceable role in speech science research and 
technical applications such as speech recognition and synthesis and speech analysis. Speech corpus 
plays an important role in corpus-based speech synthesis or recognition system, which directly affects 
the quality and efficiency of speech synthesis or recognition. In this corpus, sound and text coexist, 
original data, transcribed data and annotated data exist at the same time, and there are all kinds of 
complicated relations between them. The media voice corpus mainly provides services for the 
teaching and scientific research of the language art of broadcasting hosts, and it has very important 
practical significance for studying the art of broadcasting hosts. 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

617 

2. Method 
2.1 Design and Development of Phonetic Corpus 

Language includes three basic attributes: form, meaning and sound. Phonetics is the most 
important attribute. Language first exists in the form of phonetics. There are languages with or 
without words in the world, but there is no language without phonetics. Phonetics plays an 
irreplaceable role in language. Although pronunciation is a kind of sound, it is essentially different 
from ordinary sounds. When we determine the pronunciation materials, according to the principle of 
pronunciation text selection, there are four parts: one is the single word part. The second part is 
vocabulary. The third part is the sentence material part. The fourth part is the natural dialogue part. 
After recording the pronunciation text, it is necessary to analyze and process the speech data to get 
the different features of the speech signal, which is the key to the design of the speech corpus and the 
necessary foundation for the later speech processing. For the recorded speech signal, the acoustic 
parameters of each segment need to be extracted. Intelligent speech technology based on speech 
synthesis, speech recognition and natural language understanding is a challenging and difficult high-
tech field. Many researchers have made a lot of research achievements in this field. 

The goal of speech synthesis is to make the computer emit speech with distinctive personality, 
natural fluency and emotion like human beings. Speech production is a process from the deep 
structure of conscious expression to the surface structure of sound wave output, including three stages: 
construction, conversion and execution, which also corresponds to the three stages of speech synthesis 
system (front-end processing, prosody analysis and synthesis), as shown in Figure 1. 

 

Figure 1. Schematic diagram of speech synthesis system 
Language Corpus is a comprehensive corpus system which faces the field of broadcasting and 

hosting art and takes radio and television news corpus as its main content. It includes two closely 
related sub-corpora, namely, text corpus and phonetic corpus. In order to ensure that the training of 
acoustic model can describe each primitive and its co-pronunciation phenomenon more accurately, 
each phonetic unit in the corpus should reach a certain coverage and balance. Generally speaking, the 
larger the primitives, the better the effect and the more the number, and the speech library will be 
very large and difficult to cover all primitives; Small primitives, strong flexibility, and more splicing 
points will lead to the degradation of speech quality. The media corpus should be representative, and 
strive to reflect the actual use of audio language in radio and television. Its main purpose is to serve 
the research of audio language in media, especially the prosody research of audio language in media. 
In practical application, the appropriate primitive model should be selected after considering the scale 
of speech database and covering all kinds of language phenomena. A good corpus construction means 
that sparse training sets are needed to enrich the feature space of as many phonetic units as possible. 
The essence of this process is how to perfect a mathematical model about coverage.The coverage 
formula is as follows: let { }NF ,...,1= , Nbe the characteristic factor number, any Fi∈∀ , define the 
specific factor number iF as a discrete set, ( )ξFFFi ,...,1= ,where iF=ξ . Define the feature space 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

618 

( )NFFFF ...21×= ,so that each feature value ρ corresponds to a specific vector. The formula is as 
follows: 

( ) ( ) ( )NifFffpf iN ≤≤∈∈= 1,,...,1                      (1) 
The addition model is another method that we consider. The eigenvalue corresponding to the 

eigenvector f is ( )fD , and the formula is as follows: 

( ) ( )∑
=

+=
N

i
ii fDDfD

1
mean                              (2) 

Where meanD is a constant, and ( ) NifD ii ≤≤=∑ 1,0 . Therefore, at least ( )∑ =
−+

N

i i1
11 ξ  sample 

is needed to solve all the parameters. 
More than 4 hours' corpus (including 14 male voices and 12 female voices) were selected from the 

media speech corpus for statistical analysis. The basic information of the selected corpus is shown in 
Table 1: 

More than 4 hours' corpus (including 14 male voices and 12 female voices) were selected from the 
media speech corpus for statistical analysis. The basic information of the selected corpus is shown in 
Table 1: 

Table 1. Specific distribution of selected annotated corpus 

Linguistic form Representative radio and television programs Temporal 
distribution 

Soliloquy 

Broadcast Point news, News broadcast, National news broadcast, etc. 1:08:56 

Explain Stories, World Weekly, China Weekly, etc. 0:45:35 

Read aloud Mainly literary works. 1:10:35 

Talk Stories,WorldWeekly,ChinaWeekly,Immediately etc. 0:50:53 

Comprehensive News horizon, etc 0:18:03 

Amount to 4:14:02 

Even if these live actual sound or background music are filtered out, the extracted voice data, 
especially the pitch data, will be greatly distorted, so it is of little significance to compare and analyze 
this kind of voice data. 

When synthesizing, the same kind of speech primitives appearing in different environments are 
processed by digital signals to adapt to different environments. With the development of waveform 
splicing system, later systems record multiple samples for each kind of phonetic primitives in the 
phonetic database. When synthesizing, choose different samples in different environments. 
According to the requirements of speech corpus for speech synthesis and speech recognition, and 
taking into account the characteristics of Chinese speech, the following criteria are put forward for 
sentence selection: 1. Select three-tone submodel as the basic unit for describing speech; 2. Covering 
all kinds of language phenomena to the maximum extent with as little corpus as possible; 3. In the 
speech synthesis and speech recognition system, the selected sentences should contain as many high-
frequency words as possible; 4. Avoid data sparseness, so that each triplet appears a certain number 
of times. The corpus selection process is shown in Figure 2. 

In this way, it is ensured that the sentences with many high-frequency words will be considered 
first when selecting the corpus. The establishment of phonetic corpus is a complicated problem, and 
there are still some problems to be improved for the later improvement of phonetic corpus, such as 
making full use of the existing inter-word tone sandhi rules to reflect the actual situation of tone 
sandhi and light tone as much as possible. For the lack of corpus, we can improve the utilization rate 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

619 

of existing corpus in preprocessing. In view of the above reasons, the voice database should be an 
open database, so that changes can be added at any time to improve the database. 

 

Figure 2. Corpus selection process 

2.2 The Art of Broadcasting and Hosting under the Background of Media 
It is more and more important to carry on the cultural function of the media. Compared with other 

types of cultural carriers, modern media has more diversified grasp of cultural characteristics, and it 
is an artistic aesthetic expression with comprehensive characteristics. In this society, everyone is an 
independent individual, and each person has his own characteristics, which is individual difference. 
In fact, the external image and the internal connotation of the broadcasting host are dialectically 
unified, and the external image is the expression and formation of the internal connotation, which is 
the solid foundation of the external image. The two support and unify each other, so it also includes 
the pronunciation ability of each person and the characteristics of his voice, which mainly refers to 
the pronunciation of each person. Television has promoted and created the spread of visual culture, 
and changed the original way of communication-print communication. Before TV media, people 
wanted to receive information, either by word of mouth or by reading books, which was inefficient 
and slow. However, the TV media has abandoned this drawback, and through the TV programs with 
sound and image signals, it spreads by video, which not only greatly improves the speed of 
information dissemination, but also greatly improves the efficiency of information dissemination. 
Thus, the artistic effects and themes of the envoys can be fully disseminated, and the artistry of the 
broadcast host can be perfectly displayed from the perspective of modern media. With the complexity 
of the media environment, the new media orientation has changed. Therefore, the broadcast host 
needs to invest more energy to analyze and study the audience market. Only by fully mobilizing the 
enthusiasm of the audience can we provide more choices for the audience under the background of 
rich choices of the current audience. Therefore, it's very important to study the artistic aesthetic 
connotation of the radio host in detail, for how to shape excellent radio host and build a successful 
radio program which is welcomed and loved by the audience. (As shown in Table 2) 

Therefore, broadcasting and hosting programs should first grasp their visual and artistic image and 
artistic connotation. In the process of the development of TV livelihood news programs, the arrival 
of the new media era, on the whole, will have more advantages than disadvantages, which will 
contribute to the development of TV livelihood news programs, and promote the artistic development 
of broadcasting and hosting programs from the perspective of modern media in combinationwith the 
comprehensiveness of current communication laws. New media technology should be adopted to 
innovate the production and dissemination of TV news about people's livelihood, so as to promote 
the better development of TV news about people's livelihood. Once the host appears in front of the 
audience, it is no longer just himself, but a general artistic image. What he conveys is not only the 
knowledge structure and life experience of the host himself, but also the values embodied in a grand 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

620 

social culture. As broadcast hosts, language is an important means for them to constantly create and 
communicate with the audience. Therefore, it is the most basic ability that broadcast hosts should 
master under the media background and one of the important elements of broadcast host art to use 
the acceptable voice of the audience and use appropriate language expression skills. 

If you want to do a good job in broadcasting, you must firmly establish the goal of serving the 
people, establish correct social values, and improve your professional skills and self-cultivation. 
Grasp the pulse of the times and express the new voice of the times. Broadcasters must have rich 
knowledge reserves to improve their sensitivity to all kinds of people when they receive languages. 
Choose a reasonable way of language application. At the same time, it is necessary to further 
strengthen the understanding of current events by broadcast hosts, so as to ensure that the materials 
they cite have strong timeliness. Under the background of media, only with good appearance, rich 
connotation, high value orientation, sound and appropriate language expression skills that can bring 
good hearing experience to the audience, can the broadcast host greatly improve the probability of 
success of the column. 

Table 2. Strategies to show the artistic charm of radio and television broadcasting hosts 

Tactics Main content 

Create a good auditory 
image 

Including the quality, accomplishment, connotation and ability of radio and 
television broadcasters. You should not only be articulate, good at communication 

and expression, but also have a keen mind and good psychological quality. You 
should be able to deal with any unexpected situation and control the radio and 

television programs freely, so as to bring the audience a pleasant hearing feeling. 

Understand the social 
value orientation 

Due to the differences in positioning, planning, organization and dissemination of 
radio and television programs, and the differences in social functions such as moral 

demonstration, government mouthpiece and social voice, radio and television 
programs put forward different social value orientations. Broadcasting art must 

realize social functions such as cognition, education, publicity and aesthetics in its 
creation, and establish its own artistic image in the social and cultural horizon. 

Familiar with language 
pronunciation skills 

In the host of radio and television programs, language should first give people a 
good aesthetic feeling in hearing, which can arouse the audience's basic 

understanding and intuitive evaluation of the cultural connotation and personality 
cultivation of radio and television hosts. 

Apply the broadcasting 
law 

The broadcasting host of radio and television stations must obey the basic laws of 
broadcasting programs of radio and television stations. In the context of a specific 

era, we should take care of the discretion and scale of the host of the radio 
program. 

3. Result analysis 
The design of the overall system structure is to divide the whole system into various functional 

modules reasonably, so as to correctly handle the relationship between modules and the inside of 
modules, as well as the data connection between them, and define the internal structure of each 
module. Includes five functional modules: user registration module, user administrator login module, 
recording module, database maintenance module and query download module. Database management 
system is a computer core subject that develops rapidly with the development of computer science 
and computer hardware in order to meet the demand of information society for data management 
technology. The main reason for choosing a database is whether it can support your application well. 
We can use the following aspects as reference factors for database selection: the convenience of 
database construction; Support for program development; Performance analysis of database; Data 
integrity constraint; Concurrency control function; Safety; Chinese character processing. A complete 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

621 

phonetic corpus needs not only original phonetic data, corresponding pronunciation texts, but also 
corresponding annotation files. To improve the utilization value of speech corpus, the key is to mark 
the speech corpus completely. The tagging process of phonetic corpus is a formalization process of 
language knowledge. The generated preliminary voice annotation file can reduce a large number of 
tedious text input work and save manpower and material resources in the process of annotation of 
voice files. Fig. 3, fig. 4, and fig. 5 show an automatically generated preliminary markup file that is 
not aligned with the voice file. 

 
Figure 3. The frequency of the first 20 Chinese characters in the original corpus 

 

 
Figure 4. The frequency of the first 20 Chinese characters in phonetic corpus 

 

 
Figure 5. The frequency of the first 20 Chinese characters in the dialogue corpus 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

622 

After screening, the corpus covers 30 common words in Chinese Common Words List and 30 
common words in Modern Chinese Common Words List, with the coverage rates of common words 
and common words reaching, and respectively. For common words and phrases that cannot be 
covered by the corpus, we will automatically generate the corpus texts of these words and phrases, 
and store them in our corpus in the form of word stock or word stock. The establishment of database 
management system in phonetic corpus needs to store four kinds of materials: one is the speaker's 
attribute materials, such as the speaker's age, gender, education, mastery of Chinese, and my use of 
mother tongue. Second, pronunciation text materials, inputting and storing the pronunciation 
materials of the speaker and the corresponding dialect pronunciation, Putonghua international 
phonetic symbols and other text materials. The third is the actual voice data material, which is mainly 
used to save the original parameters of the recorded voice waveform graphics. Fourth, acoustic 
analysis parameter data, that is, preservation of acoustic parameters extracted from processed speech 
waveform. At the same time, in the process of spreading TV program information, the broadcast host 
should learn to make good use of his own artistic conditions, and combine the overall style and 
program characteristics of TV programs, so as to form a language art with strong influence and own 
personality. The improvement of radio and television director's ability in the new period requires 
directors to keep pace with the times and constantly improve their director's ability, innovation ability 
and comprehensive level, so that radio and television works can have higher quality and level. 

4. Conclusions 
With the rapid development of computer technology, speech recognition and speech synthesis 

technology have also made rapid progress, and more and more technologies have been applied to 
practical products. Products such as speech input system and computer-assisted language learning 
system are constantly emerging, providing superior services for people's study and life. Excellent 
speech synthesis and speech recognition system, a speech corpus with high information content and 
low redundancy is essential. It can be seen that speech corpus plays an important role in speech 
recognition, speech synthesis and speech research in other fields. Broadcasting is an important part 
of TV media, which has a very important influence on the style and orientation of TV programs. Its 
artistic image, aesthetic connotation and appeal have an important influence on the media effect. In 
the following work, we can further enlarge the scope and information of the original corpus, so that 
the selected corpus text can cover more phonetic phenomena. It is also possible to further expand the 
size of the screened corpus text, so as to expand the information content of the whole speech corpus. 
Only in this way can the broadcasting host accelerate the improvement of the social influence and 
penetration of TV media in the declining status of TV media and the fierce competition of multimedia 
technologies. More importantly, it reduces the dependence of media language corpus on third-party 
tools, and makes a big step towards the goal of a completely independent media language corpus 
management system. 

In the aspect of voice recording and voice database management system, the workflow and design 
ideas of the recording module are described in detail, and the voice database management system is 
realized, which facilitates users to inquire and download corpus texts, voice files and annotation files, 
and tests the recording work and establishes a preliminary voice database. Language corpus is a 
comprehensive corpus which faces the field of broadcasting and hosting art and features the news 
corpus of radio and television. The first phase of its text corpus project has been completed and 
completed. With the continuous development of the Internet, the analysis of the artistic elements of 
broadcasting and hosting based on the network background is an important way to improve the quality 
of current broadcasting and hosting programs. 

 



BCP Social Sciences & Humanities ECSS 2022 
Volume 19 (2022)  
 

623 

References 
[1] Song Jingjing. A Comparative Study of Chinese Style Vocabulary Based on Media Corpus. Journal of 

Changsha University, vol.no.6, pp.4,2015. 
[2] Yuan Jiahong. Collection, processing and research of large-scale phonetic corpus. Linguistic Research, 

vol.no.1, pp9,2017. 
[3] Jian Xu, Reyiman Tursun, hushur islam. Research and implementation of online multilingual speech 

corpus platform. Information and Communication, vol.no.4, pp.4,2018. 
[4] Yang Jian, Li Haiguang, Zhang Xiaoling. A Study on the Construction of Bai Phonetic Corpus. Journal 

of Dali University, vol. 2, no.2, pp.6,2017. 
[5] Zhang Jielan. The innovation of radio and television broadcasting hosting art in the new era. New Media 

Research, vol.2, no.004, pp.159-160,2016. 
[6] Li Xiaoxue. Exploration of the innovative space of broadcasting hosting art in the new era. Popular 

Literature and Art: Academic Edition, vol.no.2, pp.1,2016. 
[7] Liang Nan. On the Art of Broadcast Hosting-Making Broadcast Hosting more personalized, personalized 

and friendly. Think Tank Times, vol.no.12, pp.2,2017. 
[8] Wu Huiping. Innovative discussion on the art of radio and television broadcasting in the new era. Western 

Radio and Television, vol.no.13, pp.1,2016. 
[9] Oqir Asna, Buyin. The innovation of radio and television broadcasting hosting art in the new era. Science 

and Technology Communication, vol.no.15, pp.2,2016. 
[10] Dark Dream Qi. Innovation Space of Broadcasting and Hosting Art. Educational Science: Full Text 

Edition, vol.000, no.003, pp.00202-00202,2016. 
[11] Lu Ming. Innovative research on the art of radio and television broadcasting in the new era. Information 

Weekly, vol.no.31, pp.1,2019. 
[12] Cheng Yong. Discussion on the elements of broadcasting hosting art from the perspective of media. New 

Media Research, vol.no.6, pp.2,2016. 
[13] Nie Yuanfang. Exploration of innovative space of broadcasting hosting art in the new era. News Research 

Guide, vol. 8, no.11, pp.2,2017. 
[14] Du Qianqian. On the innovation of radio and television broadcasting hosting art in the new era. Western 

Radio and Television, vol.no.17, pp.1,2015. 
[15] Xu Jiaxin. Summarize the innovation and future development of broadcasting hosting art. Naked, 

vol.no.12, pp.1,2015. 


	1. Introduction
	2. Method
	2.1 Design and Development of Phonetic Corpus
	2.2 The Art of Broadcasting and Hosting under the Background of Media

	3. Result analysis
	4. Conclusions
	References