France Telecom

Orange

Delphine Charlet

Jean-Luc Grimault

Thierry Morel

Pascal Le Mer

Sylvie Laniepce

Thibaut Feuillet

Marc Denjean

Antoine Gonot

Yassine Mami

A method of navigating in a sound content wherein at least one key word is stored in association with at least two positions representative of said key word in the sound content, and wherein the method comprises: a step of displaying a representation of the sound content; during playback of the sound content, a step of detecting a current extract representative of a key word stored at a first position; a step of determining at least one second extract representative of said key word and a second position as a function of the stored positions; and a step of highlighting the position of the extracts in the representation of the sound content. The invention also relates to a system adapted to implement the navigation method.

Method of navigating in a sound content

For analyzing vocal signals of a speaker, a probability density representing resemblances between a vocal representation of the speaker in a predetermined model and a predetermined set of vocal representations of a number E of reference speakers in said predetermined model is used. The probability density is analyzed so as to deduce information on the vocal signals.

Method and system for analysis of vocal signals for a compressed representation of speakers using a probability density representing resemblances between a vocal representation of the speaker in a predetermined model and a predetermined set of vocal representations reference speakers

During a learning phase, a speech recognition device generates parameters of an acceptance voice model relating to a voice segment spoken by an authorized speaker and a rejection voice model. It uses normalization parameters to normalize a speaker verification score depending on the likelihood ratio of a voice segment to be tested and the acceptance model and rejection model. The speaker obtains access to a service application only if the normalized score is above a threshold. According to the invention, a module updates the normalization parameters as a function of the verification score on each voice segment test only if the normalized score is above a second threshold.

Verification score normalization in a speaker voice recognition device

A contributor who is one of a number of participants divided into groups at telephone terminals is identified by recognizing the contributor voice during a teleconference managed by a conference device connected to the terminals. After the voice of the contributor received by the conference device is identified, an identifier of the identified contributor is broadcast from the conference device to the terminals and reproduced at the terminals. The contributor is introduced to the other participants, automatically or on demand, for example by a voice or visual message including the contributor name. The conference device can be included in one of the terminals; for example, the identifier identifies a contributor other than the regular user of a communicating telephone or mobile radio telephone.

Identification of participant in a teleconference

A method of identifying a caller with a telephone service operator. Of the invention, said method comprises the following steps: for the operator, drawing up a list of literal expressions and allocating the caller an identifier consisting of a combination of consecutive literal expressions taken from each of said lists, for the caller, connecting to said telephone service and speaking, in order, each literal expression of said combination, and for the operator, identifying the caller by using voice recognition to search for the combination of literal expressions allocated to the caller. Application to secure access to telephone services.

Growing community of inventors

Lannion, France

Delphine Charlet