{ "owner": "facebookresearch", "repo": "fairseq", "number": 5142, "html_url": "https://github.com/facebookresearch/fairseq/issues/5142", "is_pull_request": false, "state": "closed", "state_reason": "completed", "title": "text after filtering OOV is empty output ", "author": "lisea2017", "created_at": "2023-05-24T03:02:51Z", "updated_at": "2023-05-25T01:20:41Z", "closed_at": "2023-05-25T01:20:41Z", "labels": [ "bug", "needs triage" ], "milestone": null, "comments_count": 8, "reactions": { "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }, "resolution_days": 0.93, "fix": { "closing_commit": null, "linked_prs": [], "best_guess_fix_commit": null, "has_fix": false }, "fetched_at": "2026-07-28T12:22:20.290020+00:00", "comments": [ { "author": "chevalierNoir", "created_at": "2023-05-24T04:50:46Z", "body": "@lisea2017 `jvn` is \"Javanese Surinamese\", not Japanese. " }, { "author": "lisea2017", "created_at": "2023-05-24T06:34:35Z", "body": "Sorry, if I want to try Japanese, which one should I use?" }, { "author": "w-ww", "created_at": "2023-05-24T06:58:20Z", "body": "There is no TTS model supporting Japanese. You can check the list.\r\nhttps://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html" }, { "author": "lisea2017", "created_at": "2023-05-24T07:04:03Z", "body": "If I want to support Japanese, is there a plan and what should I do?" }, { "author": "serkandyck", "created_at": "2023-05-24T14:36:57Z", "body": "> There is no TTS model supporting Japanese. You can check the list. https://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html\r\n\r\nsame on turkish oov filtering letter like \"\u00c7\", \"\u0130\", \"\u00d6\"\r\n![image](https://github.com/facebookresearch/fairseq/assets/12444059/f914edcd-39ca-4283-a137-fa087d8eb473)\r\n" }, { "author": "chevalierNoir", "created_at": "2023-05-24T16:08:49Z", "body": "> > There is no TTS model supporting Japanese. You can check the list. https://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html\r\n> \r\n> same on turkish oov filtering letter like \"\u00c7\", \"\u0130\", \"\u00d6\" ![image](https://user-images.githubusercontent.com/12444059/240640022-f914edcd-39ca-4283-a137-fa087d8eb473.png)\r\n\r\nSee the thread [here](https://github.com/facebookresearch/fairseq/issues/5125#issuecomment-1561477259) " }, { "author": "chevalierNoir", "created_at": "2023-05-24T16:35:12Z", "body": "> If I want to support Japanese, is there a plan and what should I do?\r\n\r\n@lisea2017 Japanese TTS is not supported due to some technical issues. But you could train a TTS model for Japanese using the VITS code with some public Japanese speech datasets." }, { "author": "lisea2017", "created_at": "2023-05-25T01:20:41Z", "body": "ok.thanks" } ] }