Noise Cleaning Technology of Speech Signals
Noise cleaner technology can be regarded as a separate area in speech technologies.
In some cases, poor quality of the resulting record creates certain difficulties for the information decryption. The reasons for this are failed or inept choice and usage of transmission and record medium of audio information and objective difficulties and even impossibility of obtaining high-quality “clean” audio recording in certain specific circumstances.
While processing the record technician faces the following challenges:
- Carry out maximum possible speech cleaning in order to remove noise and distortions
- Examine sound recordings
For any operation with low-quality audio recordings there is need for technically competent trained professionals and the various technical means.
During the examination it is usually required to establish occurrence of traces of deliberate assembly or recording backup; determine the type or to identify a specific instance of recording device; state the recording circumstances, method, environmental conditions, location of sound recording facilities, type of noise source, and so on; establish identity or voice difference in this (disputed) recording with voice which sample is in another (relative) audio record. In some cases, it’s required to determine speaker’s distinctive features in the speech recording (examples, gender, age, place of birth, occupation, diseases, etc.).
Facilities and technologies of noise cleaner ensure bringing quality and intelligibility of the speech signal to a level acceptable to its comfortable listening and understanding preserving important properties and characteristics of the speaker and circumstances of obtaining information, which is a very important factor.
Noise cleaner technology of speech signals is classified by type of distortion:
BABBLE NOISE - crowd noise cancellation technology
MUSIC NOISE - music noise cancellation technology
NONSTATIONARY NOISE - transient noise cancellation technology
PULSE NOISE - impulsive noise cancellation technology
TONAL NOISE - tonal noise cancellation technology
WIDEBAND NOISE - broadband noise cancellation technology
GSM HINDRANCE - noise pickup cancellation technology of cellular telephones
CLIPPING - signal clipping
REVERBERATION & NOISE – reverberation
Speech sound recording is widespread as a security tools technique in various areas of human activity. In particular, verbal directions and negotiations of air crews and large vehicles, personnel of complex technical objects, telephone conversations of various medical, emergency and law-enforcement services, commercial organizations are recorded. In some cases the fixation of speech and of other acoustic signals in industrial areas is required. Explicit and hidden means of recording are used for logging of operational information during negotiations, meetings, lectures and accident prevention actions. In some areas of human activity precise deciphering of recording speech is essential.
The noise cleaner technology is used in the products: Sound Cleaner http://speetech.com/
ANF (Automatic Noise Filter) STC-H156 http://speetech.com/
Denoiser Box http://speetech.com/
# # #
Speech Technology Ltd. designs and develops science intensive software for recording, processing and analysis of speech data. Our products make it possible to recognize speech, change voice and speech rate, implement voice biometrics, analyze audio/video data and many others. Our solutions are warning systems, stenographing systems, audio editors, video surveillance systems and many others. Our team of developers is professionals in the range of implementation of cross-platform applications from mobile devices to server stations. Detailed information is on our web site: http://speetech.com/