An end-to-end speech recognizer based on CTC structure

Abstract

The present invention relates to the technical field of speech recognition devices, and discloses an end-to-end speech recognition device based on a CTC structure, including a mounting plate, a shell is fixedly connected to the top of the mounting plate, a plurality of sound holes are provided on the top of the shell, a fixing seat is fixedly connected to the top inner wall of the shell, a second slide groove is provided at the bottom of the fixing seat, a slide seat is movably connected in the second slide groove, a fixing frame is fixedly connected to the top inner wall of the shell, a motor is fixedly connected to the bottom outer wall of the fixing frame, and one end of the motor output shaft passes through the fixing frame and is fixedly connected to a disc. The present invention can not only clean the sound holes through the coordinated use of a scraper and a torsion spring, thereby improving the cleaning effect of the device, but also can realize end-to-end transmission with the Internet through the coordinated use of an MCU module and a speech recognition module, thereby improving the intelligence of the device, and can also clean the heat dissipation holes through a scraper, thereby improving the scraping effect of the device.

Description

Technical Field

[0001] The present invention relates to the technical field of speech recognizers, and in particular to an end-to-end speech recognizer based on a CTC structure. Background Art

[0002] With the advancement of data processing technology and the rapid popularization of mobile Internet, computer technology has been widely used in various fields of society, and with it comes the generation of massive data. Among them, voice data has received more and more attention. In recent years, speech recognition technology has made significant progress and has begun to move from the laboratory to the market. Speech recognition technology is an important branch of artificial intelligence, involving many disciplines, and is a key link in human-computer natural interaction technology. Speech recognition technology is widely used, including end-to-end speech recognizers that use CTC network communication technology.

[0003] Speech recognizers are generally equipped with sound holes to facilitate the electrical components inside the speech recognizer to perform speech recognition processing. However, after long-term use, the sound holes are easily clogged with dust, which has a certain impact on speech recognition. Therefore, it is very necessary to propose an end-to-end speech recognizer based on the CTC structure. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides an end-to-end speech recognizer based on a CTC structure, which is mainly used to solve the problem that after long-term use of the speech recognizer, the sound holes are easily contaminated with dust, which has a certain impact on speech recognition.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A CTC-based end-to-end speech recognizer comprises a mounting plate, wherein the top of the mounting plate is fixedly connected to a shell, a plurality of sound-transmitting holes are provided on the top of the shell, a fixing seat is fixedly connected to the top inner wall of the shell, a second slide groove is provided on the bottom of the fixing seat, a slide seat is movably connected in the second slide groove, a fixing frame is fixedly connected to the top inner wall of the shell, a motor is fixedly connected to the bottom outer wall of the fixing frame, one end of the motor output shaft passes through the fixing frame and is fixedly connected to a disc, a connecting rod is movably connected to the top of the disc, and the connecting rod is rotatably connected to the slide seat, a rotating shaft is movably connected to the top inner wall of the shell, two scrapers are fixedly connected to one side of the rotating shaft, a torsion spring is movably connected to one side of the rotating shaft, and two ends of the torsion spring are respectively fixed to the rotating shaft and the shell, a reel is fixedly connected to one side of the rotating shaft, a pull rope is wound around one side of the reel, and one end of the pull rope is fixed to the slide seat, and an electrical component fixing plate is fixedly connected to the top of the mounting plate.

[0009] Furthermore, the top of the electrical component fixing plate is fixedly connected with an MCU module, a SIM card slot, a voice recognition module, a microphone and a speaker, and the MCU module, the SIM card slot, the voice recognition module, the microphone and the speaker are electrically connected.

[0010] Based on the above scheme, multiple heat dissipation holes are opened on both sides of the shell, and first slide grooves are opened on the inner walls of both sides of the shell. A scraper frame is movably connected in the first slide groove, and a connecting plate is fixedly connected to one side of the scraper frame, and the connecting plate is fixed to the slide seat.

[0011] As a further solution of the present invention, the top inner wall of the shell is movably connected to two connecting shafts, one side of the connecting shaft is fixedly connected to an auxiliary blade, one side of the connecting shaft is fixedly connected to a toothed disk, one side of the connecting plate is fixedly connected to a fixing rod, one side of the fixing rod is fixedly connected to a rack, and the rack is meshed with the toothed disk.

[0012] Furthermore, one side of the two fixing rods is fixedly connected with a fixing rib, and the fixing rib is fixed to the connecting plate.

[0013] Based on the above solution, a sealing gasket is fixedly connected to one side of the mounting plate.

[0014] As a further solution of the present invention, a camera is provided on the top of the shell, and a plurality of anti-slip grooves are provided on the outer walls on both sides of the shell.

[0015] Furthermore, both side outer walls of the shell are fixedly connected with fixing ears, and a plurality of fixing holes are formed at the bottom of the fixing ears.

[0016] (III) Beneficial effects

[0017] Compared with the prior art, the present invention provides an end-to-end speech recognizer based on the CTC structure, which has the following beneficial effects:

[0018] 1. Through the coordinated use of the scraper and the torsion spring, the motor rotates to drive the disc to rotate, and the scraper rotates forward and backward through the connecting rod, slide, pull rope, unwinding roller, rotating shaft and torsion spring, so that the scraper cleans the sound-transmitting hole to avoid dust adhesion to the sound-transmitting hole after long-term use, which has a certain impact on voice recognition and improves the cleaning effect of the device.

[0019] 2. Through the coordinated use of the MCU module and the voice recognition module, the microphone can collect the sounds made by people, and the voice recognition module can recognize the sounds and convert them into digital information, and then the digital information is processed and analyzed by the MCU module. At the same time, the SIM card slot is equipped with a SIM network communication card, so that the analyzed digital information can be transmitted end-to-end with the Internet through the SIM network communication card, and then the speaker can give voice prompts, which improves the intelligence of the device.

[0020] 3. By installing a scraper in the first slide groove, the slide seat moves left and right so that the connecting plate drives the scraper to move left and right along the first slide groove, so that the heat dissipation holes are cleaned by the scraper, avoiding the heat dissipation holes being blocked by dust after long-term use, affecting the heat dissipation efficiency of the voice recognizer, and improving the scraping effect of the device.

[0021] 4. Through the coordinated use of the auxiliary blades and the connecting shaft, the connecting plate moves left and right to make the fixed rod drive the rack to move left and right, and the rack moves left and right to make the gear plate drive the connecting shaft and the auxiliary blades to rotate forward and reverse, so that the auxiliary blades disturb the air in the outer shell, increase the rate at which the air inside the outer shell and the air outside the outer shell flow through the heat dissipation holes, and improve the heat dissipation effect of the device.

[0022] 5. By installing fixing ribs on the fixing rod, the fixing ribs can make the connection between the fixing rod and the connecting plate more stable, avoiding the breakage of the connection between the fixing rod and the connecting plate after long-term use, thereby improving the stability of the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the front side stereoscopic structure of an end-to-end speech recognizer based on a CTC structure proposed by the present invention;

[0024] Figure 2 A schematic diagram of the bottom three-dimensional structure of an end-to-end speech recognizer based on a CTC structure proposed by the present invention;

[0025] Figure 3 A schematic diagram of the internal structure of an end-to-end speech recognizer based on a CTC structure proposed by the present invention;

[0026] Figure 4 A schematic diagram of the decomposition structure of an end-to-end speech recognizer based on a CTC structure proposed in the present invention;

[0027] Figure 5 A schematic diagram of a partial cross-sectional structure of an end-to-end speech recognizer based on a CTC structure proposed by the present invention;

[0028] Figure 6 A schematic diagram of the enlarged structure of the A part of an end-to-end speech recognizer based on the CTC structure proposed by the present invention;

[0029] Figure 7 This is a schematic diagram of the enlarged structure of part B of an end-to-end speech recognizer based on a CTC structure proposed by the present invention;

[0030] Figure 8 This is a schematic diagram of the C-part amplification structure of an end-to-end speech recognizer based on the CTC structure proposed by the present invention.

[0031] In the figure: 1. camera; 2. sound-transmitting hole; 3. shell; 4. fixing hole; 5. fixing ear; 6. anti-slip pattern; 7. heat dissipation hole; 8. sealing gasket; 9. mounting plate; 10. scraper; 11. electrical component fixing plate; 12. MCU module; 13. SIM card slot; 14. voice recognition module; 15. microphone; 16. speaker; 17. pull rope; 18. scraper; 19. first slide groove; 20. disc; 21. fixing frame; 22. connecting plate; 23. unwinding roller; 24. rotating shaft; 25. torsion spring; 26. slide seat; 27. connecting rod; 28. second slide groove; 29. ​​fixing seat; 30. fixing rib; 31. fixing rod; 32. toothed disc; 33. rack; 34. connecting shaft; 35. auxiliary blade. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Reference Figure 1-Figure 8A CTC structure-based end-to-end speech recognizer comprises a mounting plate 9, a housing 3 is fixed to the top of the mounting plate 9 by bolts, a plurality of sound-transmitting holes 2 are provided on the top of the housing 3, a fixing seat 29 is fixed to the top inner wall of the housing 3 by bolts, a second slide groove 28 is provided at the bottom of the fixing seat 29, a slide seat 26 is slidably connected in the second slide groove 28, a fixing frame 21 is fixed to the top inner wall of the housing 3 by bolts, a motor is fixed to the bottom outer wall of the fixing frame 21 by bolts, and one end of the motor output shaft passes through the fixing frame 21 and is fixed to a disc 20 by bolts, a connecting rod 27 is rotatably connected to the top of the disc 20, and the connecting rod 27 is rotatably connected to the slide seat 26, a rotating shaft 24 is rotatably connected to the top inner wall of the housing 3, two scrapers 10 are fixed to one side of the rotating shaft 24 by bolts, a torsion spring 25 is sleeved on one side of the rotating shaft 24, and the two ends of the torsion spring 25 are respectively fixed to the rotating shaft 24 and the housing 3, and the rotating A reel 23 is fixed to one side of the shaft 24 by bolts, and a pull rope 17 is wound around one side of the reel 23, and one end of the pull rope 17 is fixed to the slide 26. The top of the mounting plate 9 is fixed to the electrical component fixing plate 11 by bolts. When it is necessary to clean the sound hole 2, the motor is started first, and the rotation of the motor drives the disc 20 to rotate. The rotation of the disc 20 causes the connecting rod 27 to drive the slide 26 to move left and right along the second slide groove 28. The slide 26 moves to the right, causing the pull rope 17 to drive the rotating shaft 24 to rotate through the reel 23. The rotation of the rotating shaft 24 causes the torsion spring 25 to deform and generate torsion. At the same time, the rotation of the rotating shaft 24 drives the scraper 10 to rotate. When the slide 26 moves to the left, the torsion spring 25 rebounds and resets, causing the rotating shaft 24 to drive the scraper 10 to rotate in the opposite direction and reset. This reciprocating cycle allows the scraper 10 to clean the sound hole 2 to avoid dust from sticking to the sound hole 2 after long-term use, which has a certain impact on voice recognition.

[0034] In particular, in the present invention, the top of the electrical component fixing plate 11 is fixed with an MCU module 12, a SIM card slot 13, a voice recognition module 14, a microphone 15 and a speaker 16 by bolts, and the MCU module 12, the SIM card slot 13, the voice recognition module 14, the microphone 15 and the speaker 16 are electrically connected. The microphone 15 can collect the sounds made by people, and recognize the sounds made by the voice recognition module 14 and convert them into digital information, and then the digital information is processed and analyzed by the MCU module 12. At the same time, the SIM card slot 13 is installed with a SIM network communication card, so that the analyzed digital information can be transmitted end-to-end with the Internet through the SIM network communication card, and then the speaker 16 can give voice prompts.

[0035] It should be noted that a plurality of heat dissipation holes 7 are provided on both sides of the housing 3, and a first slide groove 19 is provided on the inner wall of both sides of the housing 3, and a scraper 18 is slidably connected in the first slide groove 19, and a connecting plate 22 is fixed to one side of the scraper 18 by bolts, and the connecting plate 22 is fixed to the slide seat 26, and the heat dissipation holes 7 can assist the voice recognizer in heat dissipation. At the same time, the slide seat 26 moves left and right so that the connecting plate 22 drives the scraper 18 to move left and right along the first slide groove 19, so that the heat dissipation holes 7 are cleaned by the scraper 18 to avoid long-term After long-term use, the heat dissipation holes 7 are clogged with dust, which affects the heat dissipation efficiency of the speech recognizer. The top inner wall of the housing 3 is rotatably connected to two connecting shafts 34. An auxiliary blade 35 is fixed to one side of the connecting shaft 34 by bolts. A toothed disc 32 is keyed to one side of the connecting shaft 34. A fixing rod 31 is fixed to one side of the connecting plate 22 by bolts. A rack 33 is fixed to one side of the fixing rod 31 by bolts, and the rack 33 is meshed with the toothed disc 32. At the same time, the connecting plate 22 moves left and right to make the fixing rod 31 drive the rack 33 to move left and right. The rack 33 moves left and right so that the toothed disc 32 drives the connecting shaft 34 and the auxiliary blades 35 to rotate forward and backward, so that the auxiliary blades 35 disturb the air in the shell 3, and increase the rate at which the air inside the shell 3 and the air outside the shell 3 flow through the heat dissipation holes 7. One side of the two fixing rods 31 is fixed with fixing ribs 30 by bolts, and the fixing ribs 30 are fixed to the connecting plate 22. The fixing ribs 30 can make the connection between the fixing rod 31 and the connecting plate 22 more stable, avoiding the situation that the connection between the fixing rod 31 and the connecting plate 22 is broken after long-term use. A sealing gasket 8 is bonded to one side of the mounting plate 9, and the sealing gasket 8 can make the mounting plate 9 fit more closely with the shell 3. A camera 1 is provided on the top of the shell 3, and the camera 1 can perform real-time monitoring. A plurality of anti-skid patterns 6 are provided on the outer walls on both sides of the shell 3, and the anti-skid patterns 6 can increase the friction of the shell 3. The outer walls on both sides of the shell 3 are fixed with fixing ears 5 by bolts, and a plurality of fixing holes 4 are provided at the bottom of the fixing ears 5. The fixing holes 4 can facilitate the staff to install and fix the voice recognizer.

[0036] Working principle of this embodiment: when it is necessary to clean the sound-transmitting hole 2, first start the motor, the motor rotation drives the disc 20 to rotate, the disc 20 rotates to make the connecting rod 27 drive the slide 26 to move left and right along the second slide groove 28, the slide 26 moves to the right to make the pull rope 17 drive the rotating shaft 24 to rotate through the unwinding roller 23, the rotating shaft 24 rotates to make the torsion spring 25 deform and generate torque, at the same time, the rotating shaft 24 rotates to drive the scraper 10 to rotate, when the slide 26 moves to the left, the torsion spring 25 rebounds and resets to make the rotating shaft 24 drive the scraper 10 to rotate in the opposite direction and reset, and so on and so forth, so that the scraper 10 cleans the sound-transmitting hole 2, and avoids the sound-transmitting hole 2 after long-term use. 2 is attached with dust, which has a certain impact on voice recognition. At the same time, the sliding seat 26 moves left and right so that the connecting plate 22 drives the scraper 18 to move left and right along the first sliding groove 19, so that the heat dissipation hole 7 is cleaned by the scraper 18, so as to avoid the heat dissipation hole 7 being blocked by dust after long-term use, thereby affecting the heat dissipation efficiency of the voice recognizer. At the same time, the connecting plate 22 moves left and right so that the fixing rod 31 drives the rack 33 to move left and right, and the rack 33 moves left and right so that the toothed disc 32 drives the connecting shaft 34 and the auxiliary blade 35 to rotate forward and reverse, so that the auxiliary blade 35 disturbs the air in the housing 3, and increases the flow rate of the air inside the housing 3 and the air outside the housing 3 from the heat dissipation hole 7.

[0037] The electrical components mentioned in this article are all connected to an external main controller and 220V mains electricity, and the main controller can be a conventional known device for controlling a computer or the like.

[0038] In the description of this article, it should be noted that relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0039] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An end-to-end speech recognizer based on a CTC structure, comprising a mounting plate (9), characterized in that: The top of the mounting plate (9) is fixedly connected to a housing (3), a plurality of sound-transmitting holes (2) are provided on the top of the housing (3), a fixing seat (29) is fixedly connected to the inner wall of the top of the housing (3), a second slide groove (28) is provided on the bottom of the fixing seat (29), a sliding seat (26) is movably connected in the second slide groove (28), a fixing frame (21) is fixedly connected to the inner wall of the top of the housing (3), a motor is fixedly connected to the outer wall of the bottom of the fixing frame (21), one end of the motor output shaft passes through the fixing frame (21) and is fixedly connected to a disc (20), a connecting rod (27) is movably connected to the top of the disc (20), and the connecting rod (27) is rotatably connected to the slide (26), the top inner wall of the shell (3) is movably connected to a rotating shaft (24), one side of the rotating shaft (24) is fixedly connected to two scrapers (10), one side of the rotating shaft (24) is movably connected to a torsion spring (25), and the two ends of the torsion spring (25) are respectively fixed to the rotating shaft (24) and the shell (3), one side of the rotating shaft (24) is fixedly connected to a reel (23), one side of the reel (23) is wound with a pull rope (17), and one end of the pull rope (17) is fixed to the slide (26), and the top of the mounting plate (9) is fixedly connected to an electrical component fixing plate (11).

2. The end-to-end speech recognizer based on the CTC structure according to claim 1, characterized in that: The top of the electrical component fixing plate (11) is fixedly connected with an MCU module (12), a SIM card slot (13), a voice recognition module (14), a microphone (15) and a speaker (16), and the MCU module (12), the SIM card slot (13), the voice recognition module (14), the microphone (15) and the speaker (16) are electrically connected.

3. The end-to-end speech recognizer based on CTC structure according to claim 1, characterized in that: A plurality of heat dissipation holes (7) are provided on both sides of the shell (3), and a first slide groove (19) is provided on both inner walls of the shell (3). A scraper frame (18) is movably connected in the first slide groove (19), and a connecting plate (22) is fixedly connected to one side of the scraper frame (18), and the connecting plate (22) is fixed to a slide seat (26).

4. The end-to-end speech recognizer based on CTC structure according to claim 3, characterized in that: The top inner wall of the housing (3) is movably connected to two connecting shafts (34), one side of the connecting shaft (34) is fixedly connected to an auxiliary blade (35), one side of the connecting shaft (34) is fixedly connected to a toothed disc (32), one side of the connecting plate (22) is fixedly connected to a fixing rod (31), one side of the fixing rod (31) is fixedly connected to a rack (33), and the rack (33) is meshed with the toothed disc (32).

5. The end-to-end speech recognizer based on CTC structure according to claim 4, characterized in that: One side of the two fixing rods (31) is fixedly connected to a fixing rib (30), and the fixing rib (30) is fixed to the connecting plate (22).

6. The end-to-end speech recognizer based on CTC structure according to claim 1, characterized in that: A sealing gasket (8) is fixedly connected to one side of the mounting plate (9).

7. The end-to-end speech recognizer based on CTC structure according to claim 5, characterized in that: A camera (1) is provided on the top of the housing (3), and a plurality of anti-slip grooves (6) are provided on the outer walls on both sides of the housing (3).

8. The end-to-end speech recognizer based on CTC structure according to claim 7, characterized in that: The outer walls on both sides of the housing (3) are fixedly connected with fixing ears (5), and the bottom of the fixing ears (5) is provided with a plurality of fixing holes (4).

Images