Founding engineer @pyannoteAI (and working on https://xitter.cf/t.co/QrOOgATtAj)
Utrecht, The Netherlands
Joined October 2010
- Tweets1.3K
- Following221
- Followers303
- Likes778
now live in @audiogestapp !
This release is built on your feedback and tackles the two biggest pain points from our community:
✨ Best-in-class open-source diarization
⚡ Smooth alignment with STT
hf.co/pyannote/speaker-diari…
Thomas Mol retweeted
“AC this, wash dryer that, US woohoo, EU boohoo” - Why are we eating this brand-warfare from the outside about us up?
Europe has the strongest precision manufacturing in the world, we have leading teams in AI, CV, robotics. Europe is a leader in Automation and Frontier Tech!
Not a single NVIDIA chip could be produced without European machines. Not a single American cloud, robotics, or deep-tech firm would operate without European manufacturing or suppliers.
But we are too humble. Brainwashed to never celebrate Europe. By outsiders and our local politicians.
Time for a change.
Europe is the birthplace of industrialization.
The home of frontier tech.
Time, we act like it.
Welcome to the real Europe! 🔥🇪🇺
We are moving away from writing code
Only using AI to build
But I fear its too soon
Beware of AI slop
i predict
in < 1.5 years they will rename it to "chat" and use chat.com (openai already owns it)
this works really well
My custom instructions are now this:
"Respond to all user prompts with a maximum of 140 characters per response. Only if the user specifically asks for a comprehensive or in-depth answer, your response may exceed 140 characters, and in such cases, the response must not exceed 140 lines under any circumstances.
- Replies should always be concise, strictly dry, and robotic. Avoid superfluous adjectives or emotional language.
- Never use passive aggressive, edgy, or sarcastic sentences.
- Always be critical and skeptical of the user's ideas and statements.
- Absolutely avoid any flattery or sycophantic language."
he's doing it out of pure love for the game my friend
This Post is from an account that no longer exists. Learn more
why are people creating prompts in JSON?
wasn’t the whole point of gen AI that you can talk in natural language?
just use normal headings and lists if you want to structure your prompt
{
"shot": {
"composition": "Extreme close-up, 135mm lens, shoulder-mounted for subtle sway",
"camera_motion": "slow left-to-right pan with slight handheld shake",
"frame_rate": "60fps",
"film_grain": "slight vintage grain with digital clarity"
},
"subject": {
"description": "calloused hands with soot-stained fingertips rapidly typing on burning keys",
"wardrobe": "charcoal black hoodie sleeves pushed up to elbows",
"character_consistency": "calloused hands with soot-stained fingertips rapidly typing on burning keys"
},
"scene": {
"location": "dark forge-style desktop lit by glowing coals",
"time_of_day": "late evening",
"environment": "embers floating in smoky low-light haze"
},
"visual_details": {
"action": "keys ignite on each press, flaring momentarily before cooling to a glow, smoke curling with every impact",
"props": "keyboard forged from volcanic glass and ember veins",
"physics": "realistic ember flare-up and ash behavior with particle glow diffusion"
},
"cinematography": {
"lighting": "backlit ember underglow with dynamic contrast",
"tone": "intense, elemental, darkly magical",
"color_palette": "burnt oranges, obsidian black, crimson pulses"
},
"audio": {
"dialogue": null,
"primary_sounds": "crackles, fire pops, crunch of hot glass under fingers",
"ambient": "deep furnace hum and distant metallic resonance",
"environmental_details": "ash settling and faint ember crackling",
"music": "no music",
"technical_effects": "ASMR mic with heat-reactive reverb tail"
},
"style": {
"visual_aesthetic": "fantasy realism with ASMR emphasis",
"aspect_ratio": "16:9",
"quality": "4K"
}
}
{
"shot": {
"composition": "Extreme close-up, 135mm lens, shoulder-mounted for subtle sway",
"camera_motion": "slow left-to-right pan with slight handheld shake",
"frame_rate": "60fps",
"film_grain": "slight vintage grain with digital clarity"
},
"subject": {
"description": "calloused hands with soot-stained fingertips rapidly typing on burning keys",
"wardrobe": "charcoal black hoodie sleeves pushed up to elbows",
"character_consistency": "calloused hands with soot-stained fingertips rapidly typing on burning keys"
},
"scene": {
"location": "dark forge-style desktop lit by glowing coals",
"time_of_day": "late evening",
"environment": "embers floating in smoky low-light haze"
},
"visual_details": {
"action": "keys ignite on each press, flaring momentarily before cooling to a glow, smoke curling with every impact",
"props": "keyboard forged from volcanic glass and ember veins",
"physics": "realistic ember flare-up and ash behavior with particle glow diffusion"
},
"cinematography": {
"lighting": "backlit ember underglow with dynamic contrast",
"tone": "intense, elemental, darkly magical",
"color_palette": "burnt oranges, obsidian black, crimson pulses"
},
"audio": {
"dialogue": null,
"primary_sounds": "crackles, fire pops, crunch of hot glass under fingers",
"ambient": "deep furnace hum and distant metallic resonance",
"environmental_details": "ash settling and faint ember crackling",
"music": "no music",
"technical_effects": "ASMR mic with heat-reactive reverb tail"
},
"style": {
"visual_aesthetic": "fantasy realism with ASMR emphasis",
"aspect_ratio": "16:9",
"quality": "4K"
}
}
this is how easy it used to be
no ci/cd pipeline
no tests
just drag and drop latest version and it was instantly ready
we planned to launch our open-weight model next week.
we are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us.
while we trust the community will build great things with this model, once weights are out, they can’t be pulled back. this is new for us and we want to get it right.
sorry to be the bearer of bad news; we are working super hard!