47 results about "Vocal quality" patented technology

The invention discloses a training method of a voice bandwidth expansion model and a voice bandwidth expansion method, wherein the voice bandwidth expansion method comprises the following steps: acquiring to-be-expanded narrowband voice; calculating an amplitude spectrum and a phase spectrum of the to-be-expanded narrowband voice, and extracting auxiliary characteristics of the to-be-expanded narrowband voice ; processing the amplitude spectrum and the auxiliary characteristics of the to-be-expanded narrowband voice by virtue of the voice bandwidth expansion model which is obtained from training, so that the amplitude spectrum of a reconstructed bandwidth voice high-frequency band can be obtained; conducting mirror image reversing on the phase spectrum of the to-be-expanded narrowband voice in a frequency domain, and determining the phase spectrum of the bandwidth voice high-frequency band; and on the basis of the amplitude spectrum and the phase spectrum of the narrowband voice, and in combination with the amplitude spectrum and the phase spectrum of the reconstructed bandwidth voice high-frequency band, determining a bandwidth voice signal. With the application of the voice bandwidth expansion method provided by the invention, an effect of improving the tone quality and naturalness of the narrowband voice can be achieved.

The invention discloses a method for converting emotional speech by combining rhythm parameters (fundamental frequency, time length and energy) with a tone parameter (a formant), which mainly comprises the following steps of: 1, carrying out extraction and analysis of feature parameters on a Beihang University emotional speech database (BHUDES) emotional speech sample (containing neutral speech and four types of emotional speech of sadness, anger, happiness and surprise); 2, making an emotional speech conversion rule and defining each conversion constant according to the extracted feature parameters; 3, carrying out extraction of the feature parameters and fundamental tone synchronous tagging on the neutral speech to be converted; 4, setting each conversion constant according to the emotional speech conversion rule in the step 2, modifying a fundamental frequency curve, the time length and the energy and synchronously overlaying fundamental tones to synthesize a speech signal; and 5, carrying out linear predictive coding (LPC) analysis on the speech signal in the step 4 and modifying the formant by a pole of a transfer function so as to finally obtain the emotional speech rich in expressive force.

A set of toy figures each has a main body portion and a detachable brain portion that are associated with a certain archetype (e.g., policeman, skateboarder, soldier, infant, athlete, cheerleader, pirate, princess, etc). In response to a user engaging a brain portion and depressing the squishy brain, a switch activates a voice response circuit that plays a voice script stored on a main PCB assembly in the brain portion with a tonal quality associated with the archetype for the main body portion. Thus, if a brain portion matched with another archetype is attached, the assembled toy figure will say content that is not associated with the toy figure but in a manner that is appropriate. Thus, a soldier's brain would speak soldier jargon, but in an infantile way when on an infant body, providing increased variety and amusement possibilities for a set of such toy figures.

The invention discloses an earphone capable of improving conversation tone quality and method for improving conversation tone quality of the earphone. The earphone comprises a wire control structure, and a first microphone and a second microphone, arranged on the wire control structure and away from each other with a preset distance. The first microphone and the second microphone are used for respectively collecting two voice signals; the wire control structure is used for processing the two voice signals so as to filter audio data out of a preset range, and remain and strengthen the audio data in the preset range, and generating a acoustical signal after integrating all audio data and outputting to a mobile terminal, and therefore, the earphone and the method of the invention could filter the human voice and environmental noise, which are not required, and strengthen the needed conversation voice so that the user could obtain clearer conversation effect and more satisfactory and healthy conversation tone quality.

The invention provides a method for singing imitation. The method comprises the steps that corresponding audio materials of a source singer and a target singer are prepared; voice features of the source singer and voice features of the target singer are analyzed by using a STRAIGHT model; a joint GMM model of the source singer and the target singer is obtained through training by using a Gaussian mixture model; Gaussian transform functions based on inter-frame correlation are used in the tone conversion process; the tone of the source singer and the tone of the target singer are mixed in proportion; voice having the tone of the target singer is reconstructed by using the STRAIGHT model. The invention further provides a device for implementing the method. According to the method and device, the voice of the source singer can be converted into the voice having the tone of the target singer, converted tone quality is good, the tone is close to the tone of the target singer, and the tone features of the target singer can be added in proportion; especially, when the target singer is a music star, the self-confidence and interestingness of singing for the user at a digital audio and video place can be improved greatly, and meanwhile the level of the user for imitating the tone of the music star is improved.

The invention discloses a Hi-Fi tone quality identifying method based on pattern recognition. The method includes the steps: selecting a learning sample, converting a sample time domain signal into a frequency domain signal, and making a frequency attenuation mean value every 500Hz from 11050Hz to 22050Hz serve as a characteristic, wherein twenty two characteristics are provided and numbered from one to twenty two. Through statistical learning of a lot of already labeled lossy and lossless audios, a lossless audio condition probability corresponding to a lossless audio characteristic is obtained, a lossy audio condition probability corresponding to a lossy audio characteristic is obtained, Fourier transform is performed for a to-be-measured audio, and a combined probability can be calculated and then the probability of the specific tone quality of a given audio can be reckoned after condition probabilities between the characteristics and between the characteristics and the tone quality are obtained by extracting the frequency characteristics. Through the above algorithm, a high detection success rate of real lossless audios is achieved, and a relatively low misjudgement rate of false lossless audios is guaranteed. Moreover, compared with auCDtect, the method can reach the advanced level in the industry.

A voice analysis unit (13) extracts voice feature quantities such as format and pitch from an inputted voice signal. On the basis of the extracted voice feature quantities, a masking sound generation unit (14) processes sound source data (generic masking sounds) recorded in a database (15), and generates a masking sound for outputting. An example of a generic masking sound is a recording of voices of multiple people including both male and female voices, and including disturbance noise that does not have a lexical meaning (content of conversation not being comprehensible). The masking sound generation unit (14) matches the format of the disturbance noise to the format of the input voice signal format. The generated masking sound does not have a lexical meaning, and includes disturbance noise having a similar voice quality and pitch to that of a speaker; hence, it is difficult for a listener to understand what is actually being said by the speaker.

The invention provides a deep-learning-based audio tone quality enhancement method, device, and system, a storage medium and a computer program. The method comprises: acquiring lossy audio data and performing feature extraction on the lossy audio data to obtain features of the lossy audio data; and on the basis of the features of the lossy audio data, reconstructing the lossy audio data by using atrained audio reconstruction neural network to form output audio data with the tone quality close to lossless audio data. According to the invention, the lossy audio tone quality is enhanced based ondeep learning and thus the lossy audio tone quality is close to the tone quality of the lossless audio data by deep neural network reconstruction, so that the tone quality improvement effect that cannot be realized by the traditional method is realized.

The invention provides a method for monitoring and feeding back sound effect by utilizing a sensor. The method comprises the steps of sensing sound vibration data output by a special sound production device through a sensor unit, processing and computing the sound vibration data, then conducting comparison with prestored sound reference data, and when the two data are different, controlling a graphic equalizer to perform automatic regulation so that the processed and computed sound vibration data output by the special sound production device is identical to the sound reference data. Correspondingly, the invention further provides a device for monitoring and feeding back sound effect by utilizing the sensor, and a sound effect system, the best tone quality can be restored through automatic tuning so as to ensure that the tone quality is highly consistent with the original tone quality; in addition, conversion between mail voice and female voice or the effect of imitating the sound of other people can be realized by prestoring different sound reference data, or different styles of music can be showed by changing second or third harmonic of one musical instrument, therefore, the user experience can be improved effectively.

The invention discloses an in-ear structure recognition method of an earphone, a tone quality adjustment method and the earphone, and the method comprises the steps: actively playing a test audio after the earphone is successfully worn; during the test audio playing period, collecting audio signals in ears, and detecting audio gain of a low frequency band; comparing the detected audio gain with astandard audio gain: if the detected audio gain is greater than the standard audio gain, judging the in-ear structure of the earphone as an in-ear structure, and reducing the gain of a low frequency band in an audio signal output by the earphone during the normal working period of the earphone; and if the detected audio gain is less than or equal to the standard audio gain, judging the in-ear structure of the earphone to be a half in-ear structure, and improving the gain of a low frequency band in the audio signal output by the earphone during the normal working period of the earphone. According to the invention, the in-ear structure of the earphone currently worn by the user can be automatically identified, and the tone quality is adjusted according to the identified in-ear structure, sothat the user can obtain better auditory feeling.

The embodiment of the invention provides a tone quality enhancement method and a microphone. The method comprises: voice state identification is carried out on obtained sound signals, thereby identifying sound signals in a voice state and sound signals in a non-voice state among the sound signals; comfort noise conversion is carried out on the sound signals in a non-voice state, thereby generating corresponding comfort noises; and different attenuation or excitation treatment is carried out on the sound signals in a voice state based on different frequency bands, thereby generating sound signals with tone quality enhancement; and automatic gain processing is carried out on the sound signals with tone quality enhancement. Therefore, when the sound signals are processed, the good-quality sound signals are obtained, thereby improving the user experience.

The invention relates to a voice signal-based gradient boosting decision tree depression identification method. The method comprises the following steps: obtaining voice data of an interviewer and a corresponding PHQ-8 depression screening scale score, enabling a voice signal to correspond to a PHQ-8 value, selecting a training sample set for training, and testing the sample set; extracting prosodic features representing emotion and depression in the voice signal, and spectrum-based related features and tone quality features; and performing learning on the training set by adopting a machine learning method of a gradient boosting decision tree, and taking the PHQ-8 score as an output result as a basis for judging the depression degree. According to the invention, by using the gradient boosting decision tree as a learning method, the accuracy of the predicted PHQ-8 value and the timeliness of training are improved, the PHQ-8 value of the PHQ-8 depression screening scale is taken as an output result, the score of the PHQ-8 value is between 0 and 24, the depression is determined when the score is higher than 10 and lower than 20, and the severe depression is determined when the score is higher than 20. The method has higher accuracy and objectivity.

The invention discloses a non-periodic component syllable model building and speech synthesizing method and device. The method includes the steps that according to a non-periodic component representative value, of each frame of each syllable in an original speech waveform file, on each piece of frequency band information obtained through dividing, a non-periodic component spectrum fitting curve, of each syllable, on the selected frequency band information is obtained through a discrete cosine transform method, and a non-periodic component syllable model including the non-periodic component spectrum fitting curves, of all the syllables of the original speech waveform file, on the different frequency band information is generated, so that the data information, including the frequency band number *syllable frame number, in the syllable model is converted into the fitting curves including the number of frequency bands, the scale of speech model building is downsized, the system resources are saved, meanwhile, the non-periodic component spectrum fitting curve of each syllable is built, the continuity among frames of the syllables is fully considered, the original tone quality of the syllables is kept through the fitting curves, and the quality of the synthetic speech is improved in the synthesis process.

The invention discloses a Korotkoff sound quality identification and oscillography combined blood pressure measurement method and a Korotkoff sound quality identification and oscillography combined blood pressure measurement system. The method comprises the following steps of: 1) selecting a blood pressure real-time measurement way; 2) starting the blood pressure real-time monitoring by a computer; 3) calculating a blood pressure value by a blood pressure measurement optimizing algorithm mixed with Korotkoff sound quality identification and oscillography, storing and displaying the measured data; 4) monitoring the data of a remote real-time service system; and 5) analyzing and processing an intelligent signal. The system is a blood pressure and heart rate data collecting device carried by a user and connected with a data processor; and the data processor is connected to a remote medical centre service platform via the internet. The invention can be used for measurement, storage and network transmission of blood pressures in families and hospitals so as to provide long-term stable health protection services for people to be measured.

The present disclosure relates to a music string comprising precipitation hardening stainless steel. The string has a superior resistance to relaxation and is corrosion resistant, thus improving its tuning stability and maintaining its tone quality, thus prolonging its service life.

The invention discloses a voice database composition compression used for an embedded voice synthesized system and a use method. The voice database composition compression is used for an embedded system and can transform any word received into a corresponding voice and output the voice. The syllables in Chinese are taken as the basic units of the synthesized system and a voice model database; firstly an original voice model database is established based on the syllables, then the original voice model database is structurally compressed to obtain a final compressed model database. According to the method provided in the invention, the space resource taken by the synthesized system under the embedded platform can be reduced, the synthesis speed is quickened and the naturalness and phonetic quality of the synthesized voice are maintained well.

Disclosed is a double 360° sounding resonant portable audio device. The resonant portable audio device comprises a housing, and a resonant sound generator and a printed circuit board (PCB) disposed in the housing. The improvement lies in that, a sound cavity is formed in the housing, and at least three speakers which are electrically connected to the PCB circuit board are disposed in the sound cavity. The beneficial effects of the present invention are, conventional speakers are integrated in the existing resonant audio device, and a plurality of speakers is disposed in the cavity of the resonant audio device; the combination of the plurality of speakers generates a multi-directional sound effect, which is different from the directional sound effect of a single speaker. The resonant audio device combined with the plurality of speakers can achieve a double 360° sounding effect generated by the resonant 360° sounding and traditional speaker combination 360° sounding, thereby improving the sound experience of the resonant audio device greatly.

The invention provides a double-microphone earphone microphone. A circuit board is installed in a cavity, two voice tubes are installed on the circuit board, respectively an omni-directivity voice tube and a uni-directivity voice tube, a control switch can control circuits of the omni-directivity voice tube and the uni-directivity voice tube, when a switch slide sheet slides upwards, the circuit of the uni-directivity voice tube is connected, the uni-directivity voice tube starts to work, and the circuit of the omni-directivity voice tube is opened; when the switch slide sheet slides downwards, the circuit of the omni-directivity voice tube is connected, the omni-directivity voice tube starts to work, and the circuit of the uni-directivity voice tube is opened; the uni-directivity voice tube is mainly utilized during singing, voice can be accurately transmitted into the uni-directivity voice tube through a sound inlet port of a cavity upper lid, tone quality efficient transmission is achieved, noise due to voice tube air injection is reduced, the problem of poor tone quality due to the voice tube air injection when a consumer sings can be solved, and the tone quality is made more euphonic. The omni-directivity voice tube is mainly utilized in daily telephone chats, the limitation that the orientation of a sound source is not fixed can be gotten rid of, and the conversation is still clear even if body motion causes swings of the voice tube.

The invention provides a speech emotion recognition method based on spectral features and ELM. The method comprises the following steps of extracting basic characteristics of an original speech signal, wherein the basic characteristics comprise rhythm characteristics and tone quality characteristics; extracting a Mel frequency cepstrum coefficient (MFCC) and a cochlear filter cepstrum coefficient(CFCC) by using a Teager energy operator TEO algorithm, weighting the MFCC and the CFCC to obtain a teCMFCC characteristic, and fusing the teCMFCC characteristic with a basic characteristic value to construct a characteristic matrix; performing selective dimension reduction on the characteristics by using a Fisher criterion and correlation analysis, and reserving the individual characteristics ofthe speech signals; and establishing an ELM decision tree model of an extreme learning machine to finish speech emotion recognition and classification. Through the method provided by the invention, the nonlinear characteristics of the speech signals are emphasized, and good robustness is provided; the test is performed on the CASIA Chinese emotion corpus recorded by the automatization institute ofChinese academy of sciences so as to verify that the proposed speech emotion recognition algorithm based on the spectral characteristics and the ELM has good classification and recognition precisionon the Chinese speech signals.

The invention relates to a method, a system and a terminal for detecting the tone quality of earphones, wherein the method comprises the following steps: receiving acoustical signals sent by the earphones; converting the acoustical signals into a corresponding audio spectrum; and analyzing the audio spectrum with pre-set audio spectrums of different grades in a matched manner, and judging the grade of the audio spectrum according to a matched analysis result, wherein the grade of the audio spectrum corresponds to the tone quality grade of the earphones. According to the method for detecting the tone quality of the earphones, the tone quality of the earphones can be detected conveniently and rapidly; furthermore, the method is independent of other extra professional equipment; therefore, the method has high practicability and high use value; and simultaneously, the method for detecting the tone quality of the earphones can be applicable to the system and the terminal including the system.

The invention discloses a modeling method for enhancing the expressive force of a text-to-speech (TTS) system. The modeling method is characterized in that the concept of asynchronous modeling of two acoustic features namely fundamental frequency and frequency spectrum is introduced into the Trainable TTS. The method comprises the following steps: performing independent model training under the condition of the same phoneme boundary of the fundamental frequency and the frequency spectrum; and finally generating parameters respectively according to a parameter generating algorithm, and then performing TTS, thus enhancing the expressive force in synthetic speech quality.

The invention discloses a construction method of a cross-syllable Chinese speech synthesis element with a spectrum stable boundary, and belongs to the field of speech processing. Element segmentation on speech flow data starts from the center vowel of a syllable and ends at the center vowel of an adjacent syllable, and a cross-syllable element obtained from the segmentation is composed of two parts, i.e., the part of the center vowel of the final of a previous syllable and thereafter, and the part of the center vowel of the final of a current syllable and therebefore so as to obtain the cross-syllable element. According to the invention, by using such a method, coarticulation in a syllable and between syllables can be maintained, enormous splicing caused by a too short element can be prevented, the naturalness and continuity of synthetic speech are effectively improved, and the timbre performance of the synthetic speech is not affected.

The embodiment of the invention provides a high-pitch harp. The high-pitch harp comprises a first harp shaft, a second harp shaft, a third harp shaft and strings, wherein the first harp shaft, the second harp shaft and the third harp shaft are connected end to end to form a harp frame, and the external figure of the harp frame is an isosceles triangle; the first harp shaft, the second harp shaft and the third harp shaft are detachably connected; a clamping ring is arranged at one end, far away from the first string, of the second harp shaft; a retaining ring is arranged at one end, far away from the first string, of the third harp shaft; the second harp shaft and the third harp shaft are detachably connected through the clamping ring and the retaining ring; and two ends of the strings arerespectively arranged on the second harp shaft and the third harp shaft. By adopting the scheme, the high-pitch harp can accurately adjust the tension of the strings, and the tone quality of the harpis improved.

A voice quality conversion system includes: an analysis unit which analyzes sounds of plural vowels of different types to generate first vocal tract shape information for each type of the vowels; a combination unit which combines, for each type of the vowels, the first vocal tract shape information on that type of vowel and the first vocal tract shape information on a different type of vowel to generate second vocal tract shape information on that type of vowel; and a synthesis unit which (i) combines vocal tract shape information on a vowel included in input speech and the second vocal tract shape information on the same type of vowel to convert vocal tract shape information on the input speech, and (ii) generates a synthetic sound using the converted vocal tract shape information and voicing source information on the input speech to convert the voice quality of the input speech.

A method for implementing an orchestral resonance instrument with a bass extension effect is designed for solving problems that an existing national bass string instrument is slow in audio vibration, low in volume and dull in tone and a western violin is extremely inharmonious with the style of national music. A stem is designed to be in the shape of a hollow tube and is used as an auxiliary resonance cavity for a barrel which is used as a main resonance cavity, a membrane is a mocking leather membrane, at least six metal strings are uniformly distributed on the tubular stem and the barrel by means of drawing a bow, so that the main resonance cavity and the auxiliary resonance cavity realize a stereoscopic resonance effect, and audio is further filtered by the aid of a loudspeaker opening on the surface of the main resonance cavity and a loudspeaker opening on the surface of the auxiliary resonance cavity. The method for implementing the orchestral resonance instrument with a bass extension effect has the advantages that modern instrumental acoustic theories are applied on the basis of the resonance characteristics of the traditional string instrument, single structures in which a traditional bass Chinese fiddle and a Ke-hu which are used as resonance cavities are fundamentally changed, the stereoscopic resonance effect which is generated by the barrel used as the main circular resonance cavity and the tubular stem used as the auxiliary resonance cavity is highlighted, the audio of the instrument is further filtered by the aid of the loudspeaker opening of the surface of the barrel and the loudspeaker opening on the surface of the hollow stem, noise and murmur are effectively eliminated, and the tone quality, the volume and the tone are more balanced and melodious assuredly.

The invention discloses a natural horn-type mini speaker with a reversal phase structure. The mini speaker comprises a speaker body, a loudspeaker slot is arranged in the speaker body, a loudspeaker is arranged in the loudspeaker slot, and the front of the loudspeaker is towards the external of the speaker body; a natural horn type reversal phase tube is arranged in the natural horn type mini speaker with the reversal phase structure, and is a hollow tube with a spiral plane; one end of the center of the natural horn type reversal phase tube is communicated to the external of the speaker body,and a passive vibrating diaphragm is arranged at the center of the natural horn type reversal phase tube; the other end of the natural horn type reversal phase tube is communicated with the loudspeaker slot. The structure of the invention can acquire high tone quality on the mini speaker, and the unexpected technical effect is obtained.

A sounder module disclosed by the present invention comprises a module shell and a sounder monomer located in the module shell, a vibrating diaphragm of the sounder monomer divides the space in the sounder module into a front vocal cavity and a rear vocal cavity, and the sounder module is provided with a main sound outlet hole communicated with the front vocal cavity. The main sound outlet hole is formed in the side surface of the front sound cavity and is positioned on one side of the short axis of the sounder single body; the distance from the central position of the sounder single body to the opposite side wall surface of the main sound outlet hole is not less than 8.5 mm; an acoustic component for resonance sound absorption is arranged in the sounder module, the acoustic component is provided with a communicating hole in the side wall surface opposite to the main sound outlet hole, and the acoustic component is communicated with the front sound cavity through the communicating hole. By using the sounder module as a sounding device, standing trough points influencing the flatness of a sensitivity curve are eliminated, and the medium-high frequency tone quality is improved.