51 results about "Prosody" patented technology

A system and method for confirming command or speech recognition results returned by an automatic speech recognition (ASR) engine from a command issued by an operator of a vehicle or platform, such as an aircraft or unmanned air-vehicle (UAV). The operator transmits a command signal to the ASR engine, initiated by an activation means, such as a push-button (formally known as push-to-talk or push-to-recognize). A recognition result is communicated to the user and the system awaits the confirmation for a limited period of time. During this period, in one embodiment, a low tone with high prosody is played to notify the user that the system is ready to receive the confirmation. If the user quickly presses and releases the push-button a predetermined number of times (for instance, twice to make a double-click), the result is confirmed and the ASR forwards a command signal to a system controlled thereby. Otherwise, the ASR waits for another speech command.

The voice prompt of an interactive system is adjusted based upon a state of a user. An utterance of the user is received, and the state of the user is determined based upon signal processing of the utterance of the user. Once the state of the user is determined, the voice prompt is adjusted by adjusting at least one of a tone of voice of the voice prompt, a content of the voice prompt, a prosody of the voice prompt, and a gender of the voice prompt based upon the determined state of the user.

A contour for a syllable (or other speech segment) in a voice undergoing conversion is transformed. The transform of that contour is then used to identify one or more source syllable transforms in a codebook. Information regarding the context and / or linguistic features of the contour being converted can also be compared to similar information in the codebook when identifying an appropriate source transform. Once a codebook source transform is selected, an inverse transformation is performed on a corresponding codebook target transform to yield an output contour. The corresponding codebook target transform represents a target voice version of the same syllable represented by the selected codebook source transform. The output contour may be further processed to improve conversion quality.

Disclosed are novel embodiments of a speech synthesizer and speech synthesis method for generating human-like speech wherein a speech signal can be generated by concatenation from phonemes stored in a phoneme database. Wavelet transforms and interpolation between frames can be employed to effect smooth morphological fusion of adjacent phonemes in the output signal. The phonemes may have one prosody or set of prosody characteristics and one or more alternative prosodies may be created by applying prosody modification parameters to the phonemes from a differential prosody database. Preferred embodiments can provide fast, resource-efficient speech synthesis with an appealing musical or rhythmic output in a desired prosody style such as reportorial or human interest. The invention includes computer-determining a suitable prosody to apply to a portion of the text by reference to the determined semantic meaning of another portion of the text and applying the detennined prosody to the text by modification of the digitized phonemes. In this manner, prosodization can effectively be automated.

A prosody generation apparatus capable of suppressing distortion that occurs when generating prosodic patterns and therefore generating a natural prosody is provided. A prosody changing point extraction unit in this apparatus extracts a prosody changing point located at the beginning and the ending of a sentence, the beginning and the ending of a breath group, an accent position and the like. A selection rule and a transformation rule of a prosodic pattern including the prosody changing point is generated by means of a statistical or learning technique and the thus generate rules are stored in a representative prosodic pattern selection rule table and a transformation rule table beforehand. A pattern selection unit selects a representative prosodic pattern from the representative prosodic pattern selection rule table according to the selection rule. A prosody generation unit transforms the selected pattern according to the transformation rule and carries out interpolation with respect to portions other than the prosody changing points so as to generate prosody as a whole.

A language processing portion (31) analyzes a text from a dialogue processing section (20) and transforms the text to information on pronunciation and accent. A prosody generation portion (32) generates an intonation pattern according to a control signal from the dialogue processing section (20). A waveform DB (34) stores prerecorded waveform data together with pitch mark data imparted thereto. A waveform cutting portion (33) cuts desired pitch waveforms from the waveform DB (34). A phase operation portion (35) removes phase fluctuation by standardizing phase spectra of the pitch waveforms cut by the waveform cutting portion (33), and afterwards imparts phase fluctuation by diffusing only high phase components randomly according to the control signal from the dialogue processing section (20). The thus-produced pitch waveforms are placed at desired intervals and superimposed.

A method (and system) which autonomously generates a cohesive script from a text database for creating a speech corpus for concatenative text-to-speech, and more particularly, which generates cohesive scripts having fluency and natural prosody that can be used to generate compact text-to-speech recordings that cover a plurality of phonetic events.

An electronic apparatus, based on a text sentence being input, obtains prosody information of the text sentence, segments the text sentence into a plurality of sentence elements, obtains a speech in which prosody information is reflected to each of the plurality of sentence elements in parallel by inputting the plurality of sentence elements and the prosody information of the text sentence to a text to speech (TTS) module, and merges the speech for the plurality of sentence elements that are obtained in parallel to output speech for the text sentence.

A plurality of tasks of a speech synthesizing process in which at least one of speakers, emotion or situation at the time when speeches are made, and contents of the speeches is different are set (s1), word dictionaries, prosody dictionaries, and waveform dictionaries corresponding to respective tasks are organized (s2), and when a character string is to be synthesized is input with the task specified through a game system, etc., a speech synthesizing process is performed using the word dictionary, the prosody dictionary, and the waveform dictionary corresponding to the specified task (s3). Therefore, a speech message can be generated depending on the personality of a speaker, the emotion or situation at the time when a speech is made, and the contents of the speech.

Systems and methods for prosody prediction include extracting features from runtime data using a parametric model. The features from runtime data are compared with features from training data using an exemplar-based model to predict prosody of the runtime data. The features from the training data are paired with exemplars from the training data and stored on a computer readable storage medium.

Systems are configured for generating spectrogram data characterized by a voice timbre of a target speaker and a prosody style of source speaker by converting a waveform of source speaker data to phonetic posterior gram (PPG) data, extracting additional prosody features from the source speaker data, and generating a spectrogram based on the PPG data and the extracted prosody features. The systems are configured to utilize / train a machine learning model for generating spectrogram data and for training a neural text-to-speech model with the generated spectrogram data.

The present disclosure relates to a speech synthesis apparatus and method that can remove discontinuity between phoneme units when generating a synthesized sound from the phoneme units, thereby implementing natural utterances and producing a high-quality synthesized sound having stable prosody.

The invention belongs to the technical field of voice processing and image processing, and discloses an auxiliary detection method and a classifier for depression based on acoustic features and sparse mathematics, and a depression discrimination based on joint recognition of voice and facial emotion; realizing glottis through an inverse filter For signal estimation, global analysis is used for the voice signal, feature parameters are extracted, the timing and distribution characteristics of the feature parameters are analyzed, and the prosody of different emotional voices is found as the basis for emotion recognition; MFCC is used as the feature parameter to analyze the voice signal to be processed, and the Multiple sets of training data are collected from the recorded data, and a neural network model is established for discrimination; the sparse linear combination of test samples is obtained by using the sparse representation algorithm based on OMP, and the facial emotions are discriminated and classified, and the obtained results are compared with speech recognition The results are linearly combined to obtain the final probability representing each data point. The depression recognition rate has been greatly improved and the cost is low.

It would be desirable to have audio and / or video systems and processing tools that can automatically record audio / video and analyze such recordings to capture material that may be relevant to the user. In one or more embodiments disclosed herein, recordings may be compressed by using one or more tools, including but not limited to, converting speech to text and searching for relevant content, keywords, etc.; detecting and analyzing speakers and and / or classify content (e.g., events in a conversation); remove non-substantial content (e.g., silence and other extraneous content); adjust audio to increase playback speed; use rhythm and other markers in audio to identify areas of interest; Perform segmentation clustering; use pseudo-random or random samples to select content; and other methods of extracting information to provide a summary or representation of record content for user review.

The present application relates to the technical field of information processing, and discloses an audio detection method, device, electronic equipment, and a readable storage medium. The audio detection method includes: receiving the audio to be detected and the text corresponding to the audio sent by the terminal; combining the audio with the text Perform alignment processing to obtain the start and end time of each phoneme of multiple phonemes corresponding to the text in the audio; extract the phoneme feature vector of each phoneme in the audio, and obtain the audio sequence feature of the audio based on the start and end time of each phoneme; The phoneme feature vector and the audio sequence feature are used to obtain the prosody detection result of the audio; the prosody detection result includes the accent feature and the pause feature of the audio; the prosody detection result is returned to the terminal, so that the terminal displays the text corresponding to the accent feature and the pause feature. The audio detection method provided by the present application can improve the accuracy of prosody detection results.

A method (400) for representing an intended prosody in synthesized speech (152) includes receiving a text utterance (320) having at least one word (250), and selecting an utterance embedding (260) forthe text utterance. Each word in the text utterance has at least one syllable (240) and each syllable has at least one phoneme (230). The utterance embedding represents an intended prosody. For eachsyllable, using the selected utterance embedding, the method also includes: predicting a duration of the syllable by encoding linguistic features (232) of each phoneme of the syllable with a corresponding prosodic syllable embedding (245) for the syllable; predicting a pitch contour of the syllable based on the predicted duration for the syllable; and generating a plurality of fixed-length predicted pitch frames (280) based on the predicted duration for the syllable. Each fixed-length predicted pitch frame represents part of the predicted pitch contour of the syllable.

The invention provides a rhythm marking method, device and equipment. The method comprises steps that voice data of a to-be-marked text is obtained; according to the voice data, the rhythm informationof the voice data is determined, and the rhythm information is used for indicating the pause duration of the voice data; rhythm symbols of the to-be-marked text are marked according to the rhythm information of the voice data. The method is advantaged in that rhythm marking efficiency and accuracy are improved.

The disclosure provides a text generation method and device, belonging to the field of text processing. The method includes: obtaining a target text, the target text at least including target prosodic information and a stem text; invoking a rhyming text generation model constructed based on the prosody information, processing the target text, and obtaining at least one error item text, each An error item text conforms to the prosodic rule corresponding to the target prosodic information in the title stem text; according to the title stem text, the at least one error item text and the correct item text of the title stem text, determine at least one Check out the topic. By adopting the present disclosure, the efficiency of question formulation can be improved.

The present invention provides a unified training method and system for speech synthesis and speech conversion. The method includes: decoupling the coding task of speech synthesis and speech conversion into three sub-tasks, which are the extraction of content information, the extraction of speaker information and the extraction of prosody information; the content information is a language irrelevant to the speaker. information; the speaker information includes: the characteristics of the speaker; the prosody information represents how the speaker speaks the content information, reflecting the rhythm of the speech; the content information, speaker information and prosody information obtained by extraction are input into the decoding task to get the restored voice information. The solution proposed by the present invention unifies the speech synthesis and speech conversion models, avoiding the difficulty of independent construction, and improves the performance of speech synthesis and speech conversion by using unmarked speech.

The invention provides a sample marking method and device based on artificial intelligence prosody prediction. The method comprises steps that a first text sequence of unmarked prosodies corresponding to a first sample audio file is acquired; text characteristics and pronunciation duration of each character of the first text sequence are acquired; a pre-trained prosody marking model is applied to the text characteristics and the pronunciation duration of each character of the first text sequence to acquire an output mark of each character of the first text sequence; prosodic hierarchy marking for the first text sequence is carried out according to the output mark of each character of the first text sequence. The method is advantaged in that text marking cost is reduced, text marking efficiency and accuracy are improved, more training samples required for prosodic hierarchy marking can be provided, prosodic marking model performance is improved, and the voice synthesis effect is more natural.

Embodiments of the present invention provide a language prosodic boundary prediction method, device, system and storage medium. The language prosodic boundary prediction method includes extracting embedded features of the text; using each component model in at least two component models to predict the task prosodic boundary of the corresponding granularity based on the embedded features, wherein at least one component model predicts the corresponding granularity The task prosodic boundary is also based on the task prosodic boundary predicted by at least one other component model, the at least one component model is larger than the task prosodic boundary predicted by the at least one other component model; Task prosodic boundaries outside those predicted by the component models determined the final prosodic boundaries. The above technical solution unifies at least two component models respectively used to predict prosodic boundaries of tasks with different granularities to predict language prosodic boundaries under one framework, which improves the prediction effect.

Embodiments of the present invention provide a method for training a word acoustic feature system. The method includes: splicing word acoustic features output by a word acoustic feature system with a phoneme feature sequence output by a phoneme encoder to obtain a phoneme feature sequence with word acoustic features, and splicing with actual prosody features to obtain a phoneme feature sequence with prosody and word acoustics The phoneme feature sequence of the feature; adjust the coding length, add the pitch and energy features, and decode to obtain the predicted mel spectrum; train the word acoustic feature system based on the actual mel spectrum and the predicted mel spectrum. Embodiments of the present invention also provide a word acoustic feature system and a training system for the word acoustic feature system. The embodiment of the present invention uses the trained word acoustic feature system to obtain word acoustic features that not only have word meanings, but also pronunciation, and through continuous training of the word acoustic feature system, the word acoustic features are more accurate, thereby further improving speech synthesis. The quality of speech synthesis.