Speech detector circuit for a TASI system
AssigneeStorage Technology Corporation
InventorsPeter G. Ruether, John F. O'Neill
Life of this patent
Estimating the term…
Abstract
Disclosed is a centralized circuit for determining whether input signals from a plurality of channels are speech which includes memory means in which digital amplitude samples from each of the input channels are stored in real time. A control means then periodically retrieves the samples from the memory in a time faster than real time and produces a series of time compressed input signals which are compared with a variable threshold derived from a stored level for each channel. Moreover, means are disclosed for setting a new threshold whenever the time compressed input signals differs from the old level for a predetermined time, the new level thereafter being stored in place of the previously stored old level.
Claims9
A centralized circuit for determining whether input signals from a plurality of channels are speech, said circuit comprising:
- a sampling means associated with each of said input channels for continuously providing digital samples of the input signals on each of said channels;
- a memory means for storing said samples from each channel, said samples being stored at a first rate;
- a means for retrieving said samples on a selected channel at a second rate, greater than said first rate to produce a time compressed input signal for each channel;
- a first means for comparing the power of said time compressed input signal against a selected threshold whereby speech is determined to be present when said threshold is exceeded.
A centralized circuit for determining whether input signals from a plurality of channels are speech, said circuit comprising:
- a sampling means associated with each of said channels for continuously providing digital amplitude samples of the input signals on each of said channels;
- a memory means for storing said digital amplitude samples of said input signals from each channel in real time;
- a means for retrieving said samples from each one of said channels in a time faster than real time to produce a series of time compreessed input signals;
- a storage means responsive to said retrieval means for storing an old speech level for each of said channels;
- a first means for comparing the power of the time compressed input signal for each channel with the speech level threshold for that channel, said threshold being derived from said old speech level for that channel, speech being present when said threshold is exceeded.
Dependencies are inferred from the claim text.