New to me: News Analyzer that takes recordings of "The News" from my #HDHomeRun , transcribes it, breaks it into logical segments and summarizes each segment - all run offline, on local #llm

#fasterwhisper #qwen3 #ffmpeg #comskip #framework #frameworkdesktop

🔥 Người dùng faster‑whisper báo lỗi: khi chuyển lời nói có số (không phải tiền) lại xuất hiện dấu $ ngẫu nhiên. Dùng model medium nhưng vẫn khó chuẩn hoá văn bản. Gợi ý: kiểm tra cài đặt token, thêm bước post‑process (regex) để loại bỏ $ không cần, hoặc thử model large để tăng độ chính xác. #AI #fasterwhisper #transcription #CôngNghệ #Vietnam #MachineLearning

https://www.reddit.com/r/LocalLLaMA/comments/1qbrh1u/fasterwhisper_numbersdollars_accuracy_alternative/

Tôi đã xây dựng một công cụ GUI đơn giản và nhanh chóng để chuyển đổi âm thanh không phải tiếng Anh thành phụ đề tiếng Anh, hoạt động ngoại tuyến. #FasterWhisper #WhisperGUI #Phần_mềm_miễn_phí #Công_cụ_transcribe #Tiện_ích_chuyển_giọng_nói_thành_văn_bản # Dich_thuat_tu_dyn #Gui_fasterwhisper #Tien_ich_cong_cung

https://www.reddit.com/r/SideProject/comments/1p5ebb1/i_built_a_one_file_faster_whisper_gui_that/

Etaba jugando con una aplicación disponible en #flathub llamada #SpeechNote. Bajé uno de los modelos chiquitos para probar, el #FasterWhisper, y traté de hacer la transcripción de una lectura que hice en la playa del cuento #Axolotl de #JulioCortázar .

Menuda sorpresa me di viendo que el modelo tenía la capacidad transcriptora de alguien que no he leído nunca en su vida 😂 😂 :

@alextecplayz @cassidy #GoogleLens is one of the only features I miss since switching to #GrapheneOS. Very useful with no #FOSS alternatives on any platform afaik. I suppose this could be done with an extension for #GnomeShell, OCR, image classification models, & reverse image search.

#GoogleAssistant too, but there's already a decent stack with #FasterWhisper & #Wyoming with #HomeAssistant. I'd like to build a #GTK client to integrate with this on #Linux systems.

Per a recommendation from the amazing @babe, we got ourselves a mulberry silk square from a lovely Etsy seller.

Took us several attempts and guides to work out how to make it successfully work for our long hair though 😞

Last night, we finally found a video that taught us a method that works for us 🥰

How I Wrap My Wavy Hair At Night: Step By Step Breakdown

We had to leave it running and running though, as she was going through the steps too quickly for us, and we're a slow-learner when it comes to anything physical 😅

In case anyone wants or needs it, using the power of the Audio To Text (Whisper) function in Subtitle Edit, which allows you to use various engines and models, we turned the audio into text.

Here's the step-by-step breakdown of how I wrap my hair to protect it as I sleep. I'm using a large silk scarf because my hair is really long. And this is just something I had lying around my apartment. It was actually my grandma's. Starting with it over my shoulders like a cape, I'm going to tie these two front corners together, sort of like a headband, just like this. And as you can see, my hair is contained within the scarf. And then I'm going to grab the bottom corners and lift up over my head like this. So now I'm creating a little pouch for my hair. Since this scarf is so big, I then roll it down, lifting my hair up as I do this to get it up behind my head. This is what it looks like. And then I just tie these pieces at the nape of my neck. I then tuck this front piece in and under. And that's it! This is what it looks like at the end. And it's really secure so it doesn't fall off my head, but you can also Bobby pin.

For anyone else who needs to generate almost perfect transcription from audio, we highly recommend using Subtitle Edit if you want subtitles. Otherwise, check out the Whisper Github page. It gives lots of options (including links to standlone executables and programs that support the various engines.

We're using the Faster-Whisper-XXL engine and the faster-whisper-large-v3 English language model.

The Faster-Whisper executables are x86-64 compatible with Windows 7, Linux v5.4, macOS v10.15 and above, but the Faster-Whisper-XXL executables don't have macOS listed as being supported. There's probably some way to get it running though, but we're not code-savvy enough for that 😅

#Hair #LongHair #WrappingLongHair #SubtitleEdit #transcription #WhisperCCP #FasterWhisper #Python #subtitles

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

descargué un modelo para español llamado #FasterWhisper , y para el texto a voz una voz en español mexicano. Nada mal.
SOTA ASR Tooling: Long-form Transcription

Benchmarking the different Whisper frameworks for long-form transcription

Amgad’s Substack

【暫定回避策】
YukariWhisperの異常な音声認識、結果を調べると時間がかかりすぎということが分かりました。最も時間がかかるのはNG判定かな、多分そこを起点にタイムアウトを入れて5秒以上を破棄する改造をしました。
これは正直あまり意味の無いものだとは思うが、やらないよりはマシと言ったところ。本当のところは音声認識の部分に手を食われるべきなんだが、俺専門家でも職業プログラマでもないし。まあ誰かやってくれるでしょ。

#Kibekin
#ゆかりねっとコネクター
#ゆかりねっとコネクターNEO
#ゆかコネ
#ゆかNEO
#OpenAI
#Whisper
#fasterWhisper
#YukariWhisper
https://linkskibe.com/yukariwhisper_kibekin_customized_scripts_2/

評価抜ける👍ボタンはもういらないけどさ、拡散だけはお願いしてもいいかな?仕事なんでさ