5 results about "Consonant" patented technology

The application discloses an AI algorithm optimization-based sound recognition acceleration method, and particularly relates to the technical field of speech recognition, and comprises instantaneous audio frame composite feature extraction, dynamic inference configuration strategy generation, instantaneous reconstruction inference network construction, accelerated forward inference and decoding lattice generation, and optimal recognition text sequence output. The application dynamically specifies a macroscopic calculation path for a pickup input signal by calculating the instantaneous features of audio frames, and establishes shielding rules according to real-time speech speed. During inference, the network is instantaneously reconstructed according to the hierarchical regulation mechanism, shallow paths that skip the encoder layer are used for simple signals such as silence or vowels, and deep paths are called for complex consonants. The strategy double-binds the calculation resources and real-time features of signals, thereby significantly reducing the average calculation load and delay of a terminal device when processing real speech streams without sacrificing the processing capacity of complex signals.

The application discloses a Parkinson disease dysarthria early identification method based on voiceprint features and belongs to the technical field of medical diagnosis. The original speech data of a to-be-identified object is subjected to speech analysis and speech recognition, and an initial feature set related to voiceprints in a pronunciation deviation and a segment is extracted, target features with significant discriminability for Parkinson disease dysarthria identification are screened out, a dysarthria identification result of the to-be-identified object is obtained through a model, accurate vowel and consonant segment extraction and pronunciation deviation identification are realized, phoneme segment accurate cutting is realized by combining the association of a text sequence and speech data, the quantization determination of the pronunciation deviation is realized, the key dimension of the dysarthria identification is reserved through feature extraction, the early signals of Parkinson disease dysarthria are captured, the high-risk groups are marked, the early warning result is output in combination with the joint risk value calculation, the problem that the prodromal symptoms are not obvious is solved, the review interval is dynamically adjusted, and the whole-process management from identification to monitoring is realized.

This specification provides a method, apparatus, and server for processing audio data, applicable to the financial field. Based on this method, after receiving target audio data and obtaining the corresponding first data group through speech recognition, the first data group is first split according to a preset splitting rule to obtain multiple character matrices. Then, based on the pinyin data of the characters, the characters in the character matrices are mapped to corresponding letter character combinations to obtain the corresponding first letter character matrix. According to a preset exchange rule, letter character combinations in the first letter character matrix that satisfy the exchange conditions are determined. The initial consonants and / or final vowels of the letter character combinations that satisfy the exchange conditions are then exchanged to obtain the corresponding second letter character matrix. Based on the second letter character matrix, the corresponding second data group is obtained. This method can efficiently hide relevant information carried in the audio data, protecting the data security of that information.

The application discloses a kind of multi-dimension speech training plan generation method and system.The method first collects patient age, gender and speech sample, obtains seven-dimensional evaluation parameters including sound pressure, amplitude perturbation, maximum vocalization duration, fundamental frequency perturbation, vowel space area, tone impairment and speech comprehension score.Subsequently, according to the preset logic, these parameters are sequentially determined based on the parameters, and dynamically combine different training modules such as loudness, breath, pitch, vowel, glide, tone and consonant into personalized speech training plan.The application solves the problem of existing technology training scheme solidification through multi-dimensional evaluation and dynamic module matching, significantly improves the individualization degree and rehabilitation effect of speech training.