Archive: 2024 January

News
Tycho:a tookit for building high-ROI in-house speech-related services (ASR/TTS/Translation):Overview
Tutorials

Tycho: ๋†’์€ ROI ์ž์ฒด ์Œ์„ฑ ๊ด€๋ จ ์„œ๋น„์Šค(ASR/TTS/๋ฒˆ์—ญ) ๊ตฌ์ถ•์„ ์œ„ํ•œ ํˆดํ‚ท: ๊ฐœ์š”

์ž์ฒด ASR, TTS, ๋˜๋Š” ์Œ์„ฑ ๋ฒˆ์—ญ ์„œ๋น„์Šค๋ฅผ ์›ํ•˜๋Š” ๋Œ€๋ถ€๋ถ„์˜ ํŒ€์€ ๋™์ผํ•œ ๋ณต์žกํ•œ ์„ ํƒ์— ์ง๋ฉดํ•ฉ๋‹ˆ๋‹ค: ESPnet, NeMo, k2, HuggingFace์˜ ์กฐ๊ฐ๋“ค์„ ๋ถ™์ด๊ฑฐ๋‚˜, ๋ชจ๋“  ๊ฒƒ์„ ํด๋ผ์šฐ๋“œ API์— ๋„˜๊ธฐ๊ณ  ๋ถ„๋‹น ๋น„์šฉ์„ ์ง€๋ถˆ

1์›” 26, 2024ยท 1 min read
From OpenAI's Whisper Model to Your Own In-House ASR Service: Postprocessing and Language Modeling
Tutorials

OpenAI์˜ Whisper ๋ชจ๋ธ์—์„œ ์ž์‹ ์˜ ์ž์ฒด ASR ์„œ๋น„์Šค๋กœ: ํ›„์ฒ˜๋ฆฌ ๋ฐ ์–ธ์–ด ๋ชจ๋ธ๋ง

์›๋ณธ Whisper ์ถœ๋ ฅ์€ ๊ธฐ๋ณธ ์ƒํƒœ์—์„œ ์ธ์ƒ์ ์ด์ง€๋งŒ, ์‹ค์ œ ์‚ฌ์šฉ์ž์—๊ฒŒ ๋ฐฐํฌํ•œ ๋ชจ๋“  ์‚ฌ๋žŒ์€ ๋ฐ๋ชจ ํŠธ๋žœ์Šคํฌ๋ฆฝํŠธ์™€ ๋‹ค์šด์ŠคํŠธ๋ฆผ ์‹œ์Šคํ…œ์ด ์‹ค์ œ๋กœ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ๋Š” ๊ฒƒ ์‚ฌ์ด์˜ ๊ฐ„๊ฒฉ์„ ์•Œ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

1์›” 12, 2024ยท 1 min read