livellosegreto.it is one of the many independent Mastodon servers you can use to participate in the fediverse.
Livello Segreto è il social etico che ha rispetto di te e del tuo tempo.

Administered by:

Server stats:

1.2K
active users

#whisper

0 posts0 participants0 posts today

I get this “easter egg” when using whisper.cpp to transcribe audio:

```
Subtítulos realizados por la comunidad de Amara.org
0:27:19 - 0:27:22
```

That’s not part of the text, but Whisper inserts it at the end of voice transcripts. It’s baked into the AI model because OpenAI trained on movies with community-generated subtitles.

Apparently, it happens often – there's a github ticket filled with anecdotes (pictured): github.com/openai/whisper/disc

#ai#chatgpt#whisper

I've been busy with my #newapp Sussurro - Speech to Text AI 💬➡️📋 this week:

🔨 📄 1.1 Tear Down That Wall
I found a pretty ingenious way to fix #Whisper's "wall-of-text", staying true to the ethos of doing everything on device.

📺 🇩🇪 1.2 Video & German love
Sussurro now transcribes videos, again without sending data to snooping servers. And it speaks German: Der erste Markt für Sussurro ist Deutschland!

❤️ Free to try apps.apple.com/us/app/sussurro
🕵️ No tracking
💰 One-time Unlock 📱🖥️ still 50% off for 🚀

🚀 I have a #newapp!

Sussurro - Speech to Text AI 💬➡️📋

Private #Whisper text transcriptions on iPhone, iPad & Mac!

Your audio files and transcribed text are pretty sensitive: they should never leave your device, and the app you pay for should not track you.

In a category that relies too much on cloud, analytics and expensive subscriptions...

❤️ Sussurro is free to try
💪 You can test any model
🕵️ No tracking, no cloud
💰 One-time Unlock forever 50% off for launch

apps.apple.com/us/app/sussurro

App Store‎Sussurro: Speech to Text AI‎Transcribe audio to text with complete privacy, powered by the most advanced AI. Sussurro (Italian for Whisper) lets you convert audio to text using cutting-edge Artificial Intelligence models, entirely on your device. No data leaves your Mac, iPhone, or iPad — no audio uploads, no transcriptions s…

Here goes nothing: I just submitted my #newapp Sussurro, a privacy-focused #Whisper speech-to-text tool for Mac, iPhone & iPad.

So yeah, I made an hastag-AI app?

Interesting learning experience:

- 1st commit (/164) 20 days ago
- 1st #SwiftUI multi-platform project
- Making the macOS app Mac-assed enough (multi-windows, drop files) was fun
- BG downloads on iOS: challenging
- StoreKit 2 is great
- It has animations?!

and

- 1st app I made using companion AI tools, more on this in a blog post

🚀 Whisper & Pyannote: The Ultimate Combo for Speech Transcription! 🎙️

Combining Whisper (ASR) and Pyannote (diarization) enables accurate and speaker-segmented transcriptions, even locally. 🔥

💡 Applications: meetings, podcasts, sentiment analysis, subtitles...

📖 Read the article: scalastic.io/whisper-pyannote-

Have you tried these tools? Share your thoughts! 👇

Scalastic · Whisper et Pyannote : La Solution Ultime pour la Transcription de la ParoleDécouvrez Whisper et Pyannote pour transcrire la parole. Explorez les technologies de pointe en ASR et diarisation pour des retranscriptions fidèles et rapides,même en local.
#AI#ASR#Whisper

🚀 Whisper & Pyannote : la combinaison ultime pour la transcription vocale ! 🎙️

Associer Whisper (ASR) et Pyannote (diarisation) permet d’obtenir des transcriptions précises et segmentées par interlocuteur, même en local. 🔥

💡 Applications : réunions, podcasts, analyse des sentiments, sous-titres...

📖 Découvrez l’article : scalastic.io/whisper-pyannote-

Avez-vous testé ces outils ? Partagez votre avis ! 👇

Scalastic · Whisper et Pyannote : La Solution Ultime pour la Transcription de la ParoleDécouvrez Whisper et Pyannote pour transcrire la parole. Explorez les technologies de pointe en ASR et diarisation pour des retranscriptions fidèles et rapides,même en local.
#AI#ASR#Whisper

It seems some jobs' existence flies under the collective radar of society and audio transcription from interviews, public discussions and what not certainly seems to be one of these. Which means, nobody even notices that AI is taking away our whole industry, despite AI output often being far crappier as soon as someone has a dialect or uses uncommon terminology.

I appreciate people's rising awareness for accessibility in videos or podcasts or so, don't get me wrong, but who do you think are the people who transcribe audio manually? Healthy 20 year olds with no responsibilities who are capable of working full time in an office job or physical labour that pays much better than transcription?

Many people either do this additionally to something else, like translation, or are people who can only work part time for various reasons and can't just do anything else. Please don't throw us under the bus for a thing that's supposed to help others because it's so conventient to press a button in a program that burns the planet and shouts partial gibberish...

Sto dettando questo post utilizzando l'applicazione Whisper che è un'applicazione speech to text gratuita e indipendente da Google. A quanto pare il risultato è eccellente, non ha neanche bisogno della connessione a internet.

In particolare non ho dovuto correggere neanche un carattere né un segno di punteggiatura.

You can also dictate in English or in any other language. However, you should use only one language at a time.

C'est magnifique.

Potete trovare #Whisper e scaricarla da #FDroid se avete installata anche la tastiera #Heliboard, questa rileverà la presenza di Whisper e vi darà la disponibilità della dettatura.