Talker prediction method, talker prediction device, and communication system
A talker prediction method obtains a voice from a plurality of talkers, records a conversation history of the plurality of talkers, identifies a talker of the obtained voice, and predicts a next talker among the plurality of talkers based on the identified talker and the conversation history.
1. A talker prediction method comprising:
obtaining a voice from a plurality of talkers;
recording a conversation history of the plurality of talkers;
identifying a talker of the obtained voice;
detecting a part that the identified talker has talked from the conversation history; and
predicting a next talker among the plurality of talkers based on the identified talker and the conversation history,
wherein the next talker is predicted according to talk probability of a talker talking immediately after the detected part.
2. The talker prediction method according to claim 1 , further comprising controlling an image captured by a camera based on a result of the prediction.
3. The talker prediction method according to claim 1 , further comprising performing audio signal processing on an audio signal obtained by a microphone based on a result of the prediction.
4. The talker prediction method according to claim 2 , wherein the controlling the image includes framing processing.
5. The talker prediction method according to claim 3 , wherein the audio signal processing includes beamforming processing.
6. The talker prediction method according to claim 1 , wherein the talker of the obtained voice is identified based on a voice feature amount of the obtained voice.
7. The talker prediction method according to claim 1 , further comprising estimating an arrival direction of a voice, wherein the talker of the obtained voice is identified based on the arrival direction of the voice.
8. The talker prediction method according to claim 1 , further comprising obtaining an image captured by a camera, wherein the talker of the obtained voice is identified based on the image captured by the camera.
9. A talker prediction device comprising:
a voice obtainer that obtains a voice from a plurality of talkers;
a conversation history recorder that records a conversation history of the plurality of talkers;
a talker identifier that identifies a talker of the obtained voice; and
a predictor that predicts a next talker among the plurality of talkers based on the identified talker and the conversation history,
wherein the predictor detects a part that the identified talker has talked from the conversation history, and predicts the next talker according to talk probability of a talker talking immediately after the detected part.
10. The talker prediction device according to claim 9 , further comprising a camera image controller that performs control of an image captured by a camera based on a result of the prediction.
11. The talker prediction device according to claim 9 , further comprising an audio signal processor that performs audio signal processing on an audio signal obtained by a microphone based on a result of the prediction.
12. The talker prediction device according to claim 10 , wherein the control of the image includes framing processing.
13. The talker prediction device according to claim 9 , wherein the talker identifier identifies the talker of the obtained voice based on a voice feature amount of the obtained voice.
14. The talker prediction device according to claim 9 , wherein the talker identifier estimates an arrival direction of a voice, and identifies the talker of the obtained voice based on the arrival direction of the voice.
15. The talker prediction device according to claim 9 , further comprising an image obtainer that obtains an image captured by a camera, wherein the talker identifier identifies the talker of the obtained voice based on the image captured by the camera.
16. The talker prediction device according to claim 9 , wherein:
the conversation history includes respective conversation histories of a talker on a far-end side and a talker on a near-end side; and
the predictor identifies at least a voice of the talker on the far-end side and predicts a next talker on the near-end side.
17. A talker prediction method comprising:
obtaining a voice from a plurality of talkers;
recording a conversation history of the plurality of talkers;
identifying a talker of the obtained voice; and
predicting a next talker among the plurality of talkers based on the identified talker and the conversation history, wherein:
the conversation history includes respective conversation histories of a talker on a far-end side and a talker on a near-end side, and
at least a voice of the talker on the far-end side is identified to predict a next talker on the near-end side.