Design And Implementation Of System Of Robot Voice Tone

With the development of technology, audio and video resources in common has been unable to meet the demands of human society, people hope through sound entertainment funny, also hope that a bridge built between species and nature, hence the study of speech signal system has been unable to hold oneself back. Speech signal processing is a subject of study on the processing technology for digital signal processing of speech signal. It is a transmission or storage of speech signal is obtained by processing information to reflect some important features of speech parameters of speech signal in order to efficiently; two is to achieve a purpose by processing a operational requirements, such as synthesized speech, identify the speaker, identify the speech content etc..Therefore, in the study of digital speech signal processing technology is established is practical and easy to speech signal nalysis model and speech perception model. Communication with the outside life has been the bottleneck, people need to break through the phonological tone change, namely is to maintain the audio tile playback rate unchanged, according to some algorithm to adjust tone speakers,so as to realize the tone is higher or lower, to achieve the audio frequency different adjustable output will have to adjust the signal therefore, to hear all sorts of different animal sounds have to change the original speech signal fundamental frequency, and if at any time to hear the myriads of changes in the real world robot voice, to achieve happiness, anger, sorrow, joy expression with a combination of sounds, in addition to the voice tone and synthesis, With the present level of technology to achieve this level of there are still a lot of defects, in order to have a robot voice perfect, there is still a long way to go.Tone sandhi of audio processing means algorithm is a widely used, but few relevant technical materials, and introduces no detail. This paper presents a time-domain algorithm, can effectively realize the modulation constant time, and reaches the phase better continuity, experiments show good effect.Speech recognition is that has a human auditory function of the machine, can understand human speech, understand people’s intentions and respond. The input speech features compared with base speech pattern practice were obtained, thus get the recognition result. This paper mainly discusses several matching recognition algorithm, in large vocabulary continuous speech recognition system, in order to improve the recognition accuracy requires the use of language models, using speech recognition unit connection relations, the language model based method and statistical method of combining the grammar, to constrain degrees of freedom recognizer decoding, it is very difficult to use.This paper focuses on the study of phonological tone change algorithm and speech synthesis of some theory, a detailed analysis of the median filter and the linear filter phonological processing combined with removal of tonal noise effect and voice recording system, voice control system realization mechanism, adding theory of speech synthesis algorithm. The tone sandhi in practical application to robot to end the thesis, reasonable. In-depth study of the modulation algorithm, by contrast, found the time-domain modulation method is tonal a simple method, is the earliest methods change the audio frequeney.The lateral movement to achieve signal modulation, principle is: modulation of the signal in time domain, the frequency spectrum of the original signal, so as to realize the frequency increases or decreases, reaches the modulation effect. This method may cause signal distortion, tone will sound of metal and noise, but the use of linear combined with median filter can remove the big jump out of the signal, in the rough part, so that the signal close to smooth, there is a big improvement.