Audio caption generation method, audio caption generation apparatus, and program
Even in a case where an amount of training data is small, a caption for an audio signal is generated with high accuracy. An audio caption generation apparatus ( 1 ) generates a caption for an input target audio. A training data storage ( 10 ) stores a training data set including a set of an audio signal and a caption corresponding thereto. An audio similarity calculation unit ( 11 ) calculates similarity between the target audio and each audio signal of training data. A guidance caption retrieval unit ( 12 ) acquires a plurality of captions corresponding to an audio signal similar to the target audio. A caption generation unit ( 13 ) generates a caption for the target audio by determining words in order from the head on the basis of the acquired captions.
1. An audio caption generation method of generating a caption for a target audio, the audio caption generation method comprising:
causing a guidance caption retrieval circuitry to acquire a plurality of captions corresponding to an audio signal similar to the target audio; and
causing a caption generation circuitry to determine a current word of the caption for the target audio by using a feature obtained by integrating the acquired captions, a word string from the head of the caption for the target audio to an immediately preceding word of the current word, and an acoustic feature of the target audio.
2. The audio caption generation method according to claim 1 , wherein
the guidance caption retrieval circuitry is configured such that the more similar a caption for explaining a first audio signal and a caption for explaining a second audio signal is, the more likely that first audio signal and the second audio signal are determined to be similar.
3. A non-transitory computer-readable recording medium which stores a program for causing a computer to execute each step of the audio caption generation method according to claim 1 .
4. An audio caption generation apparatus that generates a caption for a target audio, the audio caption generation apparatus comprising:
a guidance caption retrieval circuitry that acquires a plurality of captions corresponding to an audio signal similar to the target audio; and
a caption generation circuitry that generates a current word of the caption for the target audio by using a feature obtained by integrating the acquired captions, a word string from the head of the caption for the target audio to an immediately preceding word of the current word, and an acoustic feature of the target audio.