Classifications

• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/28—Constructional details of speech recognition systems
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L17/00—Speaker identification or verification techniques
  • G10L17/26—Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/02—Feature extraction for speech recognition; Selection of recognition unit

Definitions

• the present invention relates to a method of recognizing a waveform by using a low-cost microcontroller.
• DSP Digital Signal Processor
• This type of electronic devices is characterized by offering great computing power, but it has the disadvantage of a relatively high cost.
• the computing power they possess is necessary to carry out the traditional method of signal analysis using the Fast Fourier Transform (FFT).
• FFT Fast Fourier Transform
• This method converts the signal from the temporal field to the frequency field, which facilitates the analysis and processing of the signals in the frequency field.
• DSP Digital Signal Processor
• This method requires that the signal to be analyzed be limited in time and repetitive. Examples of this type of signal are multiple, for example, the cry of a child, the barking of a dog, the noise of a machine, and in general all those repetitive sounds emitted by people, animals or things.
• the originality of the method resides in not using a signal analysis in the frequency field, but in the temporal field. This change, by itself, eliminates the need to use the Fast Fourier Transform and therefore the use of Digital Signal Processors.
• the method of the present patent rules out the use of patterns stored in a memory, with which the signal to be analyzed is compared. Instead, an identification procedure is performed using a fuzzy logic algorithm.
• the signal analysis method of this invention is based on the use of a low cost microcontroller that incorporates an analog / digital converter.
• the sampling phase begins.
• This converter takes a series of samples at regular intervals, of the value of the amplitude of the signal envelope.
• Direct signal samples are not taken, but the waveform of its envelope is previously obtained.
• This difference is the key to being able to use a device without great demands in terms of calculation power, since it transforms the signal to be analyzed into a similar one, but much simpler from the analysis point of view.
• the first consequence is the reduction in signal speed. This implies the possibility of using a low computing power microcontroller.
• the low frequency of the signal to be analyzed allows a much larger number of samples to be taken than the lower limit of the Nyquist frequency.
• This oversampling of the signal to be analyzed allows us to perform the analysis repeatedly with the two sample sequences.
• This repetition of the analysis allows us to compare the results obtained according to the sequence of samples analyzed and to apply different validation algorithms that guarantee the reliability of the final result.
• the way in which the waveform to be analyzed is characterized is based on the use of a matrix of temporal parameters of the waveform.
• the microprocessor performs a series of calculations to obtain the following parameters:
• a comparison is made between the elements obtained from the signal and the elements of the different reference matrices.
• the value of said index is calculated relative to the other reference matrices.
• the index does not have a prefixed value, allows us to accept waveforms with very different similarity values in appearance but that have a high degree of similarity between them in various elements of the matrix.
• the mean value could be very different from the reference value, but if the values of the rest of the elements have a high degree of correlation, the identification is positive.
• This method allows to automatically correct the decrease in the mean values of the signal as a consequence of the wear of the batteries in portable devices.
• the maximum number allowed by the internal RAM of the microcontroller (in this case 64) is taken, so the needs in terms of memory capacity are very small.
• the device in which the microcontroller has been incorporated is designed for portable use by the baby's caregiver.
• the distance from the baby's mouth to which the device should be placed is between 20 cm. And Im. T / ES2005 / 000484
• the distance limit values will depend on the regulation capacity of the block that the automatic level control performs.
• the sound is captured by means of a microphone (1) coupled to a pre-amplifier (2) that increases the level of the captured signal.
• a pre-amplifier (2) that increases the level of the captured signal.
• the necessary signal (12) is provided so that the automatic level control block (3) can supply the envelope detector (4) with the optimal signal.
• the signal envelope (13) is applied to an analog / digital converter (5) that obtains the samples of the instantaneous value of the signal (14).
• the signal identification process can be started again at any time.

Landscapes

• Engineering & Computer Science (AREA)
• Acoustics & Sound (AREA)
• Health & Medical Sciences (AREA)
• Audiology, Speech & Language Pathology (AREA)
• Human Computer Interaction (AREA)
• Physics & Mathematics (AREA)
• Multimedia (AREA)
• Computational Linguistics (AREA)
• Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
• Measurement Of Current Or Voltage (AREA)
• Feedback Control In General (AREA)
• Electrophonic Musical Instruments (AREA)
• Lock And Its Accessories (AREA)

Priority Applications (11)

Applications Claiming Priority (1)

Publications (1)

Family

ID=37835396

Family Applications (1)

Country Status (11)

Families Citing this family (4)

Citations (4)

Family Cites Families (18)

• 2005
  • 2005-09-07 AT AT05791972T patent/ATE488002T1/de not_active IP Right Cessation
  • 2005-09-07 US US12/064,988 patent/US20080284409A1/en not_active Abandoned
  • 2005-09-07 MX MX2008002313A patent/MX2008002313A/es active IP Right Grant
  • 2005-09-07 WO PCT/ES2005/000484 patent/WO2007028836A1/es not_active Ceased
  • 2005-09-07 DE DE602005024724T patent/DE602005024724D1/de not_active Expired - Lifetime
  • 2005-09-07 EP EP05791972A patent/EP1950736B1/en not_active Expired - Lifetime
  • 2005-09-07 ES ES05791972T patent/ES2354702T3/es not_active Expired - Lifetime
  • 2005-09-07 JP JP2008529649A patent/JP4931927B2/ja not_active Expired - Fee Related
  • 2005-09-07 BR BRPI0520529-8A patent/BRPI0520529A2/pt not_active IP Right Cessation
  • 2005-09-07 AU AU2005336269A patent/AU2005336269A1/en not_active Abandoned
  • 2005-09-07 CA CA002620200A patent/CA2620200A1/en not_active Abandoned

Patent Citations (4)

Also Published As

Similar Documents

Legal Events