waveglow번역#104
Conversation
JDhyeok
left a comment
There was a problem hiding this comment.
사소하지만 -합니다로 끝나는 문장에 온점이 있는 문장과 없는 문장이 혼용되고 있더라고요! 없는 문장에 온점을 추가해주시면 더 좋은 번역이 될 것 같습니다.
연휴동안 고생 많으셨습니다~~ :)
Taeyoung96
left a comment
There was a problem hiding this comment.
@poroblem
몇 가지 의견 남겼습니다. 답변 주시면 감사드리겠습니다.
HyunKyungHan
left a comment
There was a problem hiding this comment.
번역하느라 고생하셨습니다!!
몇 가지 사소한 수정 사항 제안드려요 :)
| ### Details | ||
| For detailed information on model input and output, training recipies, inference and performance visit: [github](https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/SpeechSynthesis/Tacotron2) and/or [NGC](https://ngc.nvidia.com/catalog/resources/nvidia:tacotron_2_and_waveglow_for_pytorch) | ||
| ### 세부사항 | ||
| 모델 입력 및 출력, 교육 방안, 추론 및 성과에 대한 자세한 내용은 방문하십시오 : [github](https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/SpeechSynthesis/Tacotron2) 그리고/또는 [NGC](https://ngc.nvidia.com/catalog/resources/nvidia:tacotron_2_and_waveglow_for_pytorch) |
There was a problem hiding this comment.
문장에 목적어가 없어서 조금 어색하게 느껴지는데, 방문하십시오 -> 다음 링크를 방문하십시오로 수정하는 것은 어떨까요?
| ``` | ||
|
|
||
| Load the WaveGlow model pre-trained on [LJ Speech dataset](https://keithito.com/LJ-Speech-Dataset/) | ||
| [LJ Speech datase](https://keithito.com/LJ-Speech-Dataset/)에 대해 사전 학습을 받은 WaveGlow 모델을 로드합니다. |
| body-class: hub | ||
| title: WaveGlow | ||
| summary: WaveGlow model for generating speech from mel spectrograms (generated by Tacotron2) | ||
| summary: mel spectrograms에서 발생시키기 위한 WaveGlow 모델입니다 (Tacotron2 모델에서 발생했다) |
There was a problem hiding this comment.
'speech' 해석이 빠진것 같습니다.
WaveGlow 모델은 mel spectrograms(Tacotron2 모델에서 생성)으로 만든 음성 생성을 위한 모델입니다.
로 의역해 보았습니다.
제안해 봅니다 😄
| The Tacotron 2 and WaveGlow model form a text-to-speech system that enables user to synthesise a natural sounding speech from raw transcripts without any additional prosody information. The Tacotron 2 model (also available via torch.hub) produces mel spectrograms from input text using encoder-decoder architecture. WaveGlow is a flow-based model that consumes the mel spectrograms to generate speech. | ||
|
|
||
| ### Example | ||
| Tacotron2 및 WaveGlow 모델은 사용자가 추가 운율 정보 없이 원본 텍스트에서 자연스러운 음성을 합성할 수 있는 텍스트 음성 변환 시스템을 형성합니다. Tacotron2 모델(torch.com를 통해서도 사용 가능)은 인코더-디코더 아키텍쳐를 사용하여 입력 텍스트로부터 mel spectrograms를 생성합니다. WaveGlow는 음성을 생성하기 위해 mel spectrograms를 소비하는 흐름 기반 모델입니다. |
There was a problem hiding this comment.
맨 마지막 문장
WaveGlow는 음성을 생성하기 위해 mel spectrograms를 소비하는 흐름 기반 모델입니다.
을
WaveGlow는 mel spectrograms를 사용하여 음성을 생성하는 흐름 기반 모델입니다.
로 제안해 봅니다! 😄
| ``` | ||
|
|
||
| Load a pretrained Tacotron2 model | ||
| 사전 학습을 받은 Tacotron2 모델을 로드합니다. |
There was a problem hiding this comment.
사전 학습을 받은
을
사전 학습된
으로 제안 합니다 😄
| ``` | ||
|
|
||
| Format the input using utility methods | ||
| 유용한 체계성을 사용하여 입력 형식을 지정합니다. |
There was a problem hiding this comment.
"유용한 체계성을"
을 해석하지 않고
"유틸리티 메서드를"라고 해석하는게 어떨까요?
😄
| These are needed for preprocessing the text and audio, as well as for display and input / output. | ||
| 아래의 예시에서 : | ||
| - 사전 학습을 받은 Tacotron2 및 WaveGlow 모델들은 torch.hub에서 로드됩니다. | ||
| - Tacotron2는 입력 텍스트의 텐서 표현("Hello world, I missed you so much")을 주어진 mel spectrogram을 생성합니다. |
There was a problem hiding this comment.
Tacotron2는 ("Hello world, I miss you so much")와 같은 입력 텍스트의 텐서 표현이 주어지면 그림과 같은 멜 스펙트로그램을 생성합니다.
와 같이 번역하는건 어떤가요?
해당문장에서 이해를 돕기위해 허브에서 같이 제공되는 그림을 언급하도록 원문의 구조도 변경하여 Merge한 상태입니다!.
pytorch/hub#309
| body-class: hub | ||
| title: WaveGlow | ||
| summary: WaveGlow model for generating speech from mel spectrograms (generated by Tacotron2) | ||
| summary: Tacotron2이 만들어낸 mel spectrograms을 소비하여 음성을 생성하기 위한 WaveGlow 모델입니다. |
There was a problem hiding this comment.
'Tacotron2이' 이 부분이 조금 어색합니다
만약 2를 '이'라고 읽는다면 'Tacotron2가'가 되고
'투'라고 읽는다면 역시 'Tacotron2가' 됩니다
input, ouput을 강조하기 위해 from과 for를 소비와 생성으로 바꾸었는데
조금 더 읽기 편하게 바꾸면 어떨까요?
Tacotron2가 만든 멜 스펙트로그램(mel spectrograms)으로 음성을 만드는 WaveGlow 모델입니다.
|
마지막 피드백 이후 15일이 지나면 보통 닫게 됩니다. |
라이선스 동의
변경해주시는 내용에 BSD 3항 라이선스가 적용됨을 동의해주셔야 합니다.
더 자세한 내용은 기여하기 문서를 참고해주세요.
동의하시면 아래
[ ]를[x]로 만들어주세요.관련 이슈 번호
이 Pull Request와 관련있는 이슈 번호를 적어주세요.
이슈 또는 PR 번호 앞에 #을 붙이시면 제목을 바로 확인하실 수 있습니다. (예. #999 )
PR 종류
이 PR에 해당되는 종류 앞의
[ ]을[x]로 변경해주세요.PR 설명
이 PR로 무엇이 달라지는지 대략적으로 알려주세요.