« PrécédentContinuer »
VOICE RECORD INTEGRATOR
CROSS REFERENCE TO RELATED
This application claims priority from U.S. Provisional Application No. 60/311,284, titled "Voice Record Integrator", and filed Aug. 9, 2001, the contents of which are hereby incorporated by reference.
This disclosure is directed to the use of a voice record as a signature or as a memorial of certain facts.
Every day significant business transactions take place electronically. The Census Bureau of the U.S. Department of Commerce estimates that electronic commerce sales in the 2o United States totaled $9,849 billion during the first quarter of 2002. Typically, transactions are conducted by users selecting desired products or services through a website. The user then provides payment information, such as a credit card number, and acknowledges the transaction by clicking on a button. 25 These transactions may be conducted without requiring physical or electronic signatures.
In 2000, the U.S. enacted electronic signature legislation designed to afford electronic signatures the same legal weight as written signatures. This law allows many transactions 30 required to be in writing to be executed electronically. Despite legal acceptance, it is uncommon to enter into some transactions, such as insurance agreements or real property transactions, using anything other than a written signature.
Much of electronic commerce occurs across the Internet, 35 where consumers have instant access to a plethora of information. Consumers may use an increasing variety of devices to conduct electronic commerce across the Internet such as, for example, computers, mobile phones, and personal digital assistants. For example, using a wireless access protocol 40 (WAP) enabled mobile phone, a user may browse and purchase products for delivery.
While even expensive products may be purchased electronically, some transactions usually are not entered electronically. For example, a consumer desiring an insurance 45 policy for a car, house, or boat may be able to apply for an insurance policy electronically; however, most insurance companies require that submitted information be confirmed and signed in writing before issuing the policy. Some insurance companies may issue temporary binders that terminate 50 in a short period of time unless a signed, written agreement is timely submitted by a customer.
Similarly, real estate transactions are rarely carried out electronically, though a home buyer may identify the house of their dreams through a website, contact the listing real estate 55 agent via email, and apply electronically for a mortgage. Despite the home buyer's reliance on the Internet through the whole process, the transaction closing typically involves the home buyer signing a stack of papers before a notary public.
In one general aspect, a voice record includes an electronic record, and a voice recording including a speakable electronic record identifier summarizing the electronic record. The 65 voice recording is created by taking an audio recording of a user reading a voice script including the speakable electronic
record identifier. The electronic record may include any information, such as, for example, a document, a collection of facts to be acknowledged, or an agreement. The voice recording may be created by making an audio recording of a user's voice taken using a microphone or an interactive voice response system.
In another general aspect, a method for creating voice records includes receiving information to be processed, creating a speakable identifier summarizing the received information, creating a voice script including the speakable identifier, recording a user reading the voice script, and creating a voice record including the recorded reading of the voice script. The received information to be processed may include, for example, one or more details of a transaction or a document.
Speakable identifiers may be created using any summarization technique, such as by calculating a cryptographic hash of the received information, by calculating a checksum of the received information, or by calculating a message digest of the received information.
The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features and advantages will be apparent from the description and drawings, and from the claims.
DESCRIPTION OF DRAWINGS
FIG. 1 is an exemplary network architecture for creating a voice record corresponding to electronic data.
FIG. 2 is a flowchart of a method for creating voice records corresponding to electronic data.
FIG. 3 is a diagram of the creation of a voice record by a bank teller to account for a cash transaction.
FIG. 4 is a diagram of the creation of a voice record using a speakable identifier created from a cryptographic hash.
FIG. 5 is a diagram of the creation and integration of a voice record.
FIG. 6 is a diagram of the verification of an integrated voice record.
In conducting transactions, one may be asked to sign a document such as, for example, a check, an agreement, or an application form, typically by affixing a written signature to the document. A signature is a symbol adopted with the intent to authenticate a record or document. There are no magical incantations required; it makes no difference whether a signer uses signs his name, places an arbitrary mark (e.g., an "X"), or draws a picture. The efficacy of the signature rests in the intent of the signer's action.
Just as a written signature may be affixed to a document, a voice record may be created and associated with a document. A voice record is an audio recording that may be used as a memorial of certain facts or as a signature, expressing an intent to affirm a record or transaction for legal purposes. When used as a signature, a voice record may be referred to as a voice signature. As with other types of signatures, the litmus test with voice is the ability to show that a "voice signature" was adopted with the intent to authenticate a record.
Increasingly, electronic devices (e.g., computers and personal digital assistants) are capable of recording voice statements. Analog devices such as tape recorders are also capable of recording voice statements. These devices may be used to record an individual's voice to create a voice record showing that a particular person stated particular words. These words
may indicate the facts memorialized by the voice record, or may indicate a reference to a certain document or facts.
Referring to FIG. 1, a voice record system 100 includes a workstation 102 that accesses a server 104 across a network 106. The workstation 102 is an electronic device, such as, for 5 example, a computer, a personal digital assistant, a cellular phone, a video recorder, or an audio recorder, capable of creating an audio recording for use in a voice record. The server 104 is a computer server capable of storing a voice record created using workstation 102. The workstation 102 1° and server 104 communicate across a network 106 which may be implemented using any communication technique such as, a direct cable connection, a local area network such as Ethernet, or a wide-area network such as the Internet. Using the voice record system 100, a voice record may be created to 15 represent a statement of intent or a statement of observed fact as described below.
Referring to FIG. 2, a voice record may be created by receiving information that the voice record is to memorialize (step 202). The information may be encoded as an electronic 20 document 204. The electronic document 204 may include any information such as, for example, details of a transaction, a contract, terms of an agreement, an electronic message, or the contents of a digital file. Using the information, the system creates a document identifier 208 that may be used in a voice 25 record script to represent the contents of the document or facts being memorialized.
The document identifier 210 is used to create and present a voice script for recording (step 210). The system records the user reading the voice script (step 212) to create an audio recording 214. The audio recording 214 then is used to create and store a voice record (step 216). To facilitate incorporating the document identifier 210 into a voice script, the document identifier 210 is made speakable so that it may be easily voiced by a user. One way to make a document identifier 210 speakable is to use a function to summarize the content of the document into a small, easily voiced value. For example, a small hash value may be computed that includes a small number of letters and/or numbers. In some implementations, a hash function may return a large, difficult to pronounce identifier. This identifier may be made speakable by using only a portion of that identifier, such as, for example, the first four characters of the hash or every 5th character of the hash.
Any reliable representation of the document's content may 45 be used for the document identifier 210. For example, the system may apply a hash function to document 204 to create a document identifier 210. Alternatively, the system may use any representation derived from the document's content.
Without using a document identifier, a voice record still can 50 be linked to a particular person by comparing the sound of the voice on the record to the actual voice of that person or by analyzing the circumstances under which the record was created. The circumstances may show that the particular person in question was probably the one who created the record; 55 however, the voice records without document identifier 208 may include some ambiguities such as referring to other statements or facts in a confusing way. For example, a voice record that says, "I saw them," can under the circumstances be ambiguous as to whether the speaker saw Alice and Bob, or 60 saw three articles of laundry or saw six fish.
Additionally, voice records can be moved from one storage place to another. Unless the voice record is reliably linked to the document or facts at issue, the movement can make it unclear to a future observer what the circumstances of the 65 record's creation were. Also, voice records may be ambiguous under the circumstances.
Therefore, in business transactions voice records of intent or observed facts can be improved if the records are linked or integrated more completely and unambiguously to the intent or observed facts. One way to achieve this result is to have the speaker repeat in the voice record all of his intent or all of the observed facts. For example, the speaker could state, "I now handle a red shirt. I now handle a blue sock. I now handle a green cap." But full repetition of all of the intent or facts can be tedious and can make for long and unwieldy voice records.
The implementation described with respect to FIG. 2 incorporates a document identifier 208 to strongly tie the voice record to the document 204, thus increasing the effectiveness of the resulting voice record in recording the speaker's agreement to the terms or facts described by the document 204.
In an implementation of a voice record system using the method described with respect to FIG. 2, a clothes laundering service uses voice records to record which clothes were handled by a given employee. For example, an employee, Sam, observes and handles a red shirt, a blue sock and a green cap. These are the observed facts to be acknowledged by Sam in a voice record.
First, the system receives information to be processed (step 202). In this implementation, document 204 includes a list of items of laundry obtained through any means. For example, the system may use a barcode scanner to read a barcode on each item of laundry. The collected information forms the document 204 to be acknowledged by a voice record.
The computer system then creates a document identifier (step 206) by summarizing all of the items handled by Sam. For example, the items handled by Sam may be summarized by the color of each item (i.e., "red blue green"). This summary forms the document identifier 208 which is used to create a voice script for presenting to Sam (step 210). The voice script is displayed on a computer display (or presented in any other manner) so that Sam's voice may be recorded (step 212) to create an audio recording 214 of Sam's acknowledgement of the facts. For example, the voice script may read "My name is Sam. I handled red blue green." Sam's voice is recorded using a microphone attached to the computer system. Finally, information from the document 204 and the audio recording 214 are combined to create a voice record which is stored for future verification of Sam's acknowledgment. The voice record includes a summary record integrated with many key elements of the first group of data (i.e., red shirt, blue sock, green cap) and tied to Sam's unique voice. The voice record may serve as evidence that Sam in fact handled a red shirt, a blue sock and a green cap. This would be better evidence than a voice record by Sam that simply said, "I handled three articles."
In this example, the colors of each item laundered are used to represent the information being signed or acknowledged. In other implementations, techniques such as checksums, hash functions, and message digests are used to create a document identifier. A cryptographic hash may be created using an algorithm such as RSA MD5. The algorithm yields a record, known as a hash, that is mathematically linked to the first group of data, but that is typically smaller than the first group of data. The hash serves as a control against the first group of data, such that if the first group of data is changed the hash will no longer match with the group and therefore an observer can know that the group was changed.
Various implementations may use a checksum, hash total or digest in any number of ways to facilitate the desired result. For example, a computer algorithm can summarize a first group of data by determining a checksum. A computer system can display the checksum in the form of alphanumeric characters to the speaker. The speaker then can speak into a