Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System
This paper integrates a classic mel-cepstral synthesis filter into a modern neural speech synthesis system towards end-to-end controllable speech synthesis. Since the mel-cepstral synthesis filter is explicitly embedded in neural waveform models in the proposed system, both voice characteristics and...
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | Yoshimura, Takenori Takaki, Shinji Nakamura, Kazuhiro Oura, Keiichiro Hono, Yukiya Hashimoto, Kei Nankaku, Yoshihiko Tokuda, Keiichi |
| description | This paper integrates a classic mel-cepstral synthesis filter into a modern
neural speech synthesis system towards end-to-end controllable speech
synthesis. Since the mel-cepstral synthesis filter is explicitly embedded in
neural waveform models in the proposed system, both voice characteristics and
the pitch of synthesized speech are highly controlled via a frequency warping
parameter and fundamental frequency, respectively. We implement the
mel-cepstral synthesis filter as a differentiable and GPU-friendly module to
enable the acoustic and waveform models in the proposed system to be
simultaneously optimized in an end-to-end manner. Experiments show that the
proposed system improves speech quality from a baseline system maintaining
controllability. The core PyTorch modules used in the experiments will be
publicly available on GitHub. |
| doi_str_mv | 10.48550/arxiv.2211.11222 |
| format | Article |
| fullrecord | <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2211_11222</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2211_11222</sourcerecordid><originalsourceid>FETCH-LOGICAL-a672-b573689b1e729b18b7b75ee88654b0c8a40251e734db7732c3b36156ca733ca13</originalsourceid><addsrcrecordid>eNpNj7tOwzAYhb0woMIDMOEXSIjt-MKISgtILQztiBT9dv5QS06IbIPI2xMCA8s5w7lIHyFXrCprI2V1A_HLf5acM1Yyxjk_J6-b3mLb-uGNAr33XYcRh-zBBqR7DIXDMeUIgR6mIZ8w-US3PmSMNL_Pi2f8WMIR0Z3-dQ5TythfkLMOQsLLP1-R43ZzXD8Wu5eHp_XdrgCleWGlFsrcWoaaz2qstloiGqNkbStnoK64nENRt1ZrwZ2wQjGpHGghHDCxIte_twteM0bfQ5yaH8xmwRTfFMhNAA</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System</title><source>arXiv.org</source><creator>Yoshimura, Takenori ; Takaki, Shinji ; Nakamura, Kazuhiro ; Oura, Keiichiro ; Hono, Yukiya ; Hashimoto, Kei ; Nankaku, Yoshihiko ; Tokuda, Keiichi</creator><creatorcontrib>Yoshimura, Takenori ; Takaki, Shinji ; Nakamura, Kazuhiro ; Oura, Keiichiro ; Hono, Yukiya ; Hashimoto, Kei ; Nankaku, Yoshihiko ; Tokuda, Keiichi</creatorcontrib><description>This paper integrates a classic mel-cepstral synthesis filter into a modern
neural speech synthesis system towards end-to-end controllable speech
synthesis. Since the mel-cepstral synthesis filter is explicitly embedded in
neural waveform models in the proposed system, both voice characteristics and
the pitch of synthesized speech are highly controlled via a frequency warping
parameter and fundamental frequency, respectively. We implement the
mel-cepstral synthesis filter as a differentiable and GPU-friendly module to
enable the acoustic and waveform models in the proposed system to be
simultaneously optimized in an end-to-end manner. Experiments show that the
proposed system improves speech quality from a baseline system maintaining
controllability. The core PyTorch modules used in the experiments will be
publicly available on GitHub.</description><identifier>DOI: 10.48550/arxiv.2211.11222</identifier><language>eng</language><subject>Computer Science - Computation and Language ; Computer Science - Sound</subject><creationdate>2022-11</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2211.11222$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2211.11222$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Yoshimura, Takenori</creatorcontrib><creatorcontrib>Takaki, Shinji</creatorcontrib><creatorcontrib>Nakamura, Kazuhiro</creatorcontrib><creatorcontrib>Oura, Keiichiro</creatorcontrib><creatorcontrib>Hono, Yukiya</creatorcontrib><creatorcontrib>Hashimoto, Kei</creatorcontrib><creatorcontrib>Nankaku, Yoshihiko</creatorcontrib><creatorcontrib>Tokuda, Keiichi</creatorcontrib><title>Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System</title><description>This paper integrates a classic mel-cepstral synthesis filter into a modern
neural speech synthesis system towards end-to-end controllable speech
synthesis. Since the mel-cepstral synthesis filter is explicitly embedded in
neural waveform models in the proposed system, both voice characteristics and
the pitch of synthesized speech are highly controlled via a frequency warping
parameter and fundamental frequency, respectively. We implement the
mel-cepstral synthesis filter as a differentiable and GPU-friendly module to
enable the acoustic and waveform models in the proposed system to be
simultaneously optimized in an end-to-end manner. Experiments show that the
proposed system improves speech quality from a baseline system maintaining
controllability. The core PyTorch modules used in the experiments will be
publicly available on GitHub.</description><subject>Computer Science - Computation and Language</subject><subject>Computer Science - Sound</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNpNj7tOwzAYhb0woMIDMOEXSIjt-MKISgtILQztiBT9dv5QS06IbIPI2xMCA8s5w7lIHyFXrCprI2V1A_HLf5acM1Yyxjk_J6-b3mLb-uGNAr33XYcRh-zBBqR7DIXDMeUIgR6mIZ8w-US3PmSMNL_Pi2f8WMIR0Z3-dQ5TythfkLMOQsLLP1-R43ZzXD8Wu5eHp_XdrgCleWGlFsrcWoaaz2qstloiGqNkbStnoK64nENRt1ZrwZ2wQjGpHGghHDCxIte_twteM0bfQ5yaH8xmwRTfFMhNAA</recordid><startdate>20221121</startdate><enddate>20221121</enddate><creator>Yoshimura, Takenori</creator><creator>Takaki, Shinji</creator><creator>Nakamura, Kazuhiro</creator><creator>Oura, Keiichiro</creator><creator>Hono, Yukiya</creator><creator>Hashimoto, Kei</creator><creator>Nankaku, Yoshihiko</creator><creator>Tokuda, Keiichi</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20221121</creationdate><title>Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System</title><author>Yoshimura, Takenori ; Takaki, Shinji ; Nakamura, Kazuhiro ; Oura, Keiichiro ; Hono, Yukiya ; Hashimoto, Kei ; Nankaku, Yoshihiko ; Tokuda, Keiichi</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a672-b573689b1e729b18b7b75ee88654b0c8a40251e734db7732c3b36156ca733ca13</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Computation and Language</topic><topic>Computer Science - Sound</topic><toplevel>online_resources</toplevel><creatorcontrib>Yoshimura, Takenori</creatorcontrib><creatorcontrib>Takaki, Shinji</creatorcontrib><creatorcontrib>Nakamura, Kazuhiro</creatorcontrib><creatorcontrib>Oura, Keiichiro</creatorcontrib><creatorcontrib>Hono, Yukiya</creatorcontrib><creatorcontrib>Hashimoto, Kei</creatorcontrib><creatorcontrib>Nankaku, Yoshihiko</creatorcontrib><creatorcontrib>Tokuda, Keiichi</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Yoshimura, Takenori</au><au>Takaki, Shinji</au><au>Nakamura, Kazuhiro</au><au>Oura, Keiichiro</au><au>Hono, Yukiya</au><au>Hashimoto, Kei</au><au>Nankaku, Yoshihiko</au><au>Tokuda, Keiichi</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System</atitle><date>2022-11-21</date><risdate>2022</risdate><abstract>This paper integrates a classic mel-cepstral synthesis filter into a modern
neural speech synthesis system towards end-to-end controllable speech
synthesis. Since the mel-cepstral synthesis filter is explicitly embedded in
neural waveform models in the proposed system, both voice characteristics and
the pitch of synthesized speech are highly controlled via a frequency warping
parameter and fundamental frequency, respectively. We implement the
mel-cepstral synthesis filter as a differentiable and GPU-friendly module to
enable the acoustic and waveform models in the proposed system to be
simultaneously optimized in an end-to-end manner. Experiments show that the
proposed system improves speech quality from a baseline system maintaining
controllability. The core PyTorch modules used in the experiments will be
publicly available on GitHub.</abstract><doi>10.48550/arxiv.2211.11222</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | DOI: 10.48550/arxiv.2211.11222 |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_arxiv_primary_2211_11222 |
| source | arXiv.org |
| subjects | Computer Science - Computation and Language Computer Science - Sound |
| title | Embedding a Differentiable Mel-cepstral Synthesis Filter to a Neural Speech Synthesis System |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-10T10%3A26%3A27IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Embedding%20a%20Differentiable%20Mel-cepstral%20Synthesis%20Filter%20to%20a%20Neural%20Speech%20Synthesis%20System&rft.au=Yoshimura,%20Takenori&rft.date=2022-11-21&rft_id=info:doi/10.48550/arxiv.2211.11222&rft_dat=%3Carxiv_GOX%3E2211_11222%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |