1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
|
{
"abstracts": [
{
"content": "This paper introduces Taco-VC, a novel architecture for voice conversion (VC)\nbased on the Tacotron synthesizer, which is a sequence-to-sequence with\nattention model. Most current prosody preserving VC systems suffer from target\nsimilarity and quality issues in the converted speech. To address these\nproblems, we first recover initial prosody preserving speech using a Phonetic\nPosteriorgrams (PPGs) based Tacotron synthesizer. Then, we enhance the quality\nof the converted speech using a novel speech-enhancement network, which is\nbased on a combination of phoneme recognition and Tacotron networks. The final\nconverted speech is generated by a Wavenet vocoder conditioned on Mel\nSpectrograms. Given the advantages of a single speaker Tacotron and Wavenet, we\nshow how to adapt them to other speakers with limited training data. We\nevaluate our solution on the VCC 2018 SPOKE task. Using public mid-size\ndatasets, our method outperforms the baseline and achieves competitive results",
"lang": "en",
"mimetype": "text/plain",
"sha1": "6b3ec27cabf8042bb5dd6278b9aa0cee71d92789"
}
],
"contribs": [
{
"index": 0,
"raw_name": "Roee Levy Leshem",
"role": "author"
},
{
"index": 1,
"raw_name": "Raja Giryes",
"role": "author"
}
],
"ext_ids": {
"arxiv": "1904.03522v1"
},
"extra": {
"arxiv": {
"base_id": "1904.03522",
"categories": [
"cs.SD",
"cs.LG",
"eess.AS"
],
"comments": "Submitted to Interspeech 2019"
},
"superceded": true
},
"ident": "funn7cwjbrgefji27tzpl4avuu",
"language": "en",
"license_slug": "ARXIV-1.0",
"refs": [],
"release_date": "2019-04-06",
"release_stage": "submitted",
"release_type": "article",
"release_year": 2019,
"revision": "8dd55bc1-c846-48f9-bcc6-96069460b8d8",
"state": "active",
"title": "Taco-VC: A Single Speaker Tacotron based Voice Conversion with Limited\n Data",
"version": "v1",
"work_id": "bqizapjfrfbbhnele4mba3e5ay"
}
|