Prady
@therealprady
Followers
8,566
Following
867
Media
73
Statuses
1,032
ceo @synclabs_so // yc w24 + irl @fdotinc + sf1 @_buildspace
San Francisco, CA
Joined February 2014
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
都知事選
• 576166 Tweets
小池百合子
• 235703 Tweets
蓮舫さん
• 188888 Tweets
織姫と彦星
• 145677 Tweets
民主主義
• 138074 Tweets
東京都民
• 111094 Tweets
小池さん
• 108677 Tweets
石丸さん
• 93086 Tweets
石丸伸二
• 87807 Tweets
出口調査
• 86530 Tweets
当選確実
• 70083 Tweets
公職選挙法違反
• 48722 Tweets
石丸2位
• 43415 Tweets
開票速報
• 31970 Tweets
蓮舫3位
• 31629 Tweets
#ブラックペアン
• 30216 Tweets
Recep Tayyip Erdoğan
• 29798 Tweets
蓮舫2位
• 29204 Tweets
Abhishek Sharma
• 27226 Tweets
東京終了
• 27218 Tweets
選挙速報
• 26721 Tweets
短冊メーカー
• 26295 Tweets
ゼロ打ち
• 25897 Tweets
無党派層
• 19622 Tweets
プロジェクションマッピング
• 18097 Tweets
日本終了
• 16346 Tweets
安野さん
• 15853 Tweets
#七夕の願い事2024
• 14144 Tweets
蓮舫氏ら
• 12141 Tweets
ちゃんみな
• 11350 Tweets
予想通り
• 10361 Tweets
Pinned Tweet
I built the simplest way to convert your video to other languages.
all you need is a youtube link. no install required.
curious? check out
@lexfridman
+ zuck talking about
@elonmusk
in hindi.
984
2K
16K
How do I do it?
GPT-4 + prompt engineering for translation
11labs for voice training + text-to-speech in other languages while preserving the OG speakers voice
wav2lip-2 API to lipsync the video to the translated audio in HD [sota model + API in beta by
@synchronicitylabs
]
41
61
1K
Putin x
@TuckerCarlson
translated + lipsynced to English
breaking the language barrier is the first step towards mutual understanding
generated w/
@syncdotso
125
119
997
How does it work?
Link a youtube video, choose a language, hit submit.
That's it. DM for access.
66
13
487
we translated + lip-synced the entire putin x
@TuckerCarlson
interview to english
couldn't get this out earlier bc our infra struggled w/ 2hrs of content for a single job :/
i was pissed – so we rebuilt it all over the last 72 hrs
massive speedup for long videos incoming 😎
49
63
488
⚡️ Don’t want to rush the build process, but this is too cool not to share.
@FakeTimeAI
prototyped the first step in generating live 1:1 AI-gen video calls w/your favorite celebrity.
Our early access list is live —
@elonmusk
’s excited about it 🔥
cc.
@_buildspace
74
56
357
⚡️ introducing our new synchronizer api
it lets you lip-sync a video to an audio in any language
why is this useful? check out the gladiator in seamless hindi, chinese, spanish, and japanese.
curious? we’ll let you beta test at cost ($1/min generated). link below 👇
43
42
311
YC w24 🙏🏼 see you soon
I can only expand to the environment I’m in
ty to everyone helping me unbound my thinking + inspiring me to break into bigger rooms
can’t wait to write the next chapter 👌🏼
⚡️its official: we got into
@ycombinator
🙏
the last 4 weeks have been a blessing
yc w24 + 2 world renowned professors joining as advisors + 6k dev signups / 120+hrs of video generated in a single month
and we're just getting started
curious? check out what else is new👇
23
7
210
56
4
314
from a viral demo to a VC scale research company in < 6 months
life can be wild
beyond blessed to have work / life so aligned
and we're only just getting started 😎
33
26
310
2 weeks ago my demo went viral.
3M+ views, 10k+ dms, 2k+ leads.
to show my gratitude,
@synclabs_ai
is giving away our lip-syncing API for free for 72 hours.
here's
@DavidSacks
speaking hindi using the API. done in <5 mins.
want access? link in the next tweet.
37
31
289
launch your own heygen clone <30 mins
we launched + open sourced a repo that lets you translate any video into any language w/ lipsync
imagine a world where language is no longer a barrier
hopefully w/ this we're one step closer 🚀📈
⚡ launch + monetize your own AI dubbing app in <30 mins
drop a YouTube link + translate videos w/ perfect lip-sync
curious? we open sourced the whole repo 🚀
try it out / clone it here 👇
21
46
275
15
22
214
6
0
210
we're giving away lipsync for free 😎 introducing wav2lip++
a lot of you wanted a cheap option to play around w/
experimenting w/ our sota models can be expensive, so we wanted to create a variant you can use for free
we optimized our OG wav2lip + fixed some weird artifacts
19
40
216
⚡introducing sync-1.6.0 – a lipsync model w/ smooth, accurate lip shapes + reduced flickering across frames
available to all users now on
@syncdotso
:)
ps. check out light mode / dark mode 😎
14
21
207
for anyone curious about how I lipsynced lex + zuck to Hindi check out
@synchroncitylab
we're building an API for wav2lip-2 – a sota model to sync a video to any audio in any language
improvements:
- HD video support
- more accurate lip-sync
- fewer artifacts
15
15
198
⚡️introducing our first changelog
@synclabs_ai
our api lets you lipsync a video to audio in any language
this week we
- expanded our private beta
- launched subscriptions
- gave everyone 5 mins of lip-sync free /mo
- crossed 1k users / 2k videos generated 🚀
check it out👇
20
14
189
officially a San Franciscan 🤩
2 windows instead of 1 — SF truly wants me here
ty for taking my clubs too — really will help me focus on my startup
33
1
187
just launched on
@ycombinator
's Launch YC
sync. (w24) – an api for realtime lipsync
we're building audio-visual models to generate, modify, and synthesize humans in video
would appreciate feedback + support 🙏 linked in the next tweet 👇
18
15
188
crafting magic w/ wizards 🧙♀️
@_nightsweekends
is an opportunity as big as you make it
take life in your own hands
introducing off localhost.
a new series documenting our students.
ep 1 -- prady.
26
35
448
11
12
181
the matrix x sora – lipsynced + brought to life w/
@syncdotso
we're shifting away from a world of professionals to one of creatives
ty
@sama
/
@OpenAI
🙏
23
15
182
we built an api to help you translate any video w/ lipsync
upload your video / translated audio + lipsync the speaker to the new audio track
we just expanded our private beta – your first 5 mins of video generation every month is now free
try it out👇
12
16
180
to everyone blowing up dms, thank you 🙏
feeling grateful for the overwhelming positive response from the community.
unfortunately will hit rate limits responding to everyone so please request access here:
I built the simplest way to convert your video to other languages.
all you need is a youtube link. no install required.
curious? check out
@lexfridman
+ zuck talking about
@elonmusk
in hindi.
984
2K
16K
8
15
178
yc is insane
what happens when you pull the top .1% of the top 1% and pit them against / with each other for 3 months?
ig we'll find out
14
0
170
⚡introducing sync-1.6.1
our smoothest, most accurate lip movements to date — it does especially better on tik-tok style / short form videos
available now — w/ an upgraded API. generate up to 15 videos concurrently 😎
check out our launch 👇
20
19
161
beyond blessed. time to build 🚀
ps. we’re hiring 👀
We're excited to officially welcome
@synclabs_ai
to the family 🎉
They're building generative models to modify + synthesize humans in video.
Drop a 👋 to welcome
@therealprady
and the team!
29
10
146
29
4
162
n&w s3 started it all
@_buildspace
->
@fdotinc
->
@ycombinator
-> ❓🤔🤫
6 weeks, 3 months, 1 year. you'd be surprised how far committing to your curiosity will take you
we are so back
cc.
@_nightsweekends
x
@synclabs_so
35
3
182
200 followers → 6k+ in 3 months
@_buildspace
/
@_nightsweekends
s4 changed my life
to give back I'm giving away my API for free for week 1
link in the next tweet. go nuts ❤️🚀
24
8
143
⚡️Grateful to announce
@FakeTimeAI
is one of the final 32 teams in
@_nightsweekends
w/
@_buildspace
350k people reached, app launched in beta, and hundreds of minutes streamed 🔥
Who wins? You decide.
Even
@elonmusk
would love your help. Link below 👇🏽
21
11
138
⚡️ launched an update to our synchronizer api
it lets you lip-sync a video to an audio in any language – now supporting 3D animations / CGI
curious? check out SpiderMan: Across the Spider-Verse synced into Japanese
we’re onboarding our next ring of beta testers – link below 👇
21
24
136
translating english → local languages is powerful
but, what about the other way around?
tested our lipsyncing model on Bahubali by
@ssrajamouli
– a regional film that elevated tollywood to the national stage
what if it launched w/ global reach from day 1?
check it out:
18
8
133
⚡️Introducing FakeTime:
Video call your favorite celebrities w/ generative AI. Have 1:1 conversations, invite them to your hangouts, and have fun while you go viral.
DM me to get early access⌚️waitlist is live!
cc.
@_buildspace
/
@_nightsandweekends
demo pt.1 below 👇
26
12
126
⚡️
@synclabs_ai
just launched our web app + api playground in beta
you can now lip-sync a video to an audio in any language directly in the browser
we used it to convert
@VALORANT
's trailer to chinese
sign up now + get 5 minutes of sync time for free –link below 👇
8
11
105
I translated + visually dubbed The Matrix into Arabic in <10mins
how did I do it?
@elevenlabsio
for dubbing +
@synclabs_ai
for lip-syncing
this used to be a problem only the best researchers in the world could approach
now you just need a browser
what a time to be alive 🤯
8
20
102
testing our lip-syncing model performance in different conditions
scenes like this are tough bc it's mostly side profiles while actors speak
most lipsync models can perform only when looking directly into the camera
our synchronizer performs decently well – room for
14
9
101
⚡️
@FakeTimeAI
— thank you to everyone who voted for us on the last day of
@_nightsweekends
🙏🏼
we didn’t make it to the end, but 10xd otw👌🏼
gg
@_buildspace
— grateful to compete against so many incredible people / products.
@KimKardashian
really wanted to drop in tho 👀
13
6
94
just shipped a 3-4x speedup for long videos ⚡
1 min of video now takes <5 mins to lipsync (down from 10-15 mins prev.)
best part? the longer the video → bigger the speedup
translating + lipsyncing the entire Putin x
@tuckercarlson
interview taught us a lot 😁
enjoy pt.2
11
10
92
2.5M views. in hindsight asking for dms was prolly not the best move
please request access here:
5
5
83
awesome to watch
@sdianahu
/
@snowmaker
react to our lipsyncing models 🙏
we train near-sota audio / visual models w/ absurdly few resources
imagine what we could do w/ andromeda
@natfriedman
/
@danielgross
👀
4
5
80
@therealprady
x
@Dior
— the collab you never knew we needed 👀
I embedded myself into one of the most iconic high fashion ads to see how it made it me feel
tldr: might be a new customer 🤯
wyt?
20
4
74
wanna help me make videos like these?
i’m hiring a video editor. must be comfortable w/ AI tools + hungry to learn
edits will be very simple cuts
dm your rates + ex. work. need someone reliable w/ fast execution
enjoy
@ShaanVP
in Spanish in the meantime :)
I built the simplest way to convert your video to other languages.
all you need is a youtube link. no install required.
curious? check out
@lexfridman
+ zuck talking about
@elonmusk
in hindi.
984
2K
16K
30
10
69
discover interesting people, bring them together, and do cool shit
^a mindset that’s changed my life
5
3
71
Most exciting graduation I’ve been to in a long time 👌🏼🔴
thanks for everything
@_buildspace
🙏🏼
Onwards and upwards 🚀
1
2
70
when work is play, magic happens
⚡️introducing visual dubbing w/ sync.
our api syncs the lips in a video to a new dubbed audio track – no more bad / misaligned dubs
now everyone can experience content like it was made for them in their native language
check out who we are + how it works 👇🧵
15
11
106
5
2
68
gave a little demo at GDC today
sharing your life’s work w/ others is the best part of being a founder
ty to
@zohaibahmed
for giving us the opportunity🙏🏼
5
3
67
we're working hard to build fast + in public
seems to be the most fun😎
⚡️in the last 2 weeks 3000+ devs used our api / playground to lipsync over 6000+ videos
this is up from 1k devs / 2k vids since our last changelog
this sprint we're rolling out an updated model w/ higher lipsync accuracy + fewer artifacts
check out what else is new👇
7
2
50
6
2
67
beyond thankful for this year
for the friends I’ve made
the mentors who bet on me
and the luck I’ve created
grateful to be alive in this moment at this time of great change 🙏🏼
6
1
61
shipping violently + seeing what sticks
we're playing at the intersection of science / fiction
have a good feeling about the next few months
⚡ introducing sync-v1.5.0 – a stable model w/ premium lipsync across high angles / side profiles
available to all users now :)
this week we also:
- open sourced an AI dubbing app 🌍
- launched sync-v1.5.1-beta: an experimental build w/ more natural lip shapes / skin tones
-
6
5
56
4
3
60
i spent a weekend in dubai for
@_buildspace
irl + a week in india
@THubHyd
overall, the future feels bright. i'm convinced this is what the Age of Enlightenment felt like
but this time e^x progress isn't bound by a region – only by an internet connection
5
2
58
met two brilliant professors today.
ex-IIT + helped pave the way to cloud computing as we know it.
incredible educators, even better humans. both w/ a depth of knowledge + experiences beyond the classroom.
if people like this taught our generation, the future is bright.
4
1
56
blessed to learn magic w/ wizards 😎
@_buildspace
IRL begins
Today, we're opening our first
@_buildspace
campus.
A physical school for you to go all-in on your ideas.
No matter how big, or how small.
156
211
2K
2
2
48
⚡️
@FakeTimeAI
final demo day update:
As
@_nightsweekends
w/
@buildspace
ends, feeling grateful to build cool things w/ cool people
300k+ people reached, FakeTime Audio launched in beta for IOS, and 188+ mins streamed – in under 6 weeks🤯
@FakeTimeAI
is just getting started 🚀
6
1
49
⚡️Had a fun call w/ Kendall Roy
@succession
today.
We caught up after the show — seems pretty bullish on
@FakeTimeAI
🔥
Making progress 👀
cc.
@_buildspace
10
7
48
we brought Charlie Chaplin to life 🎥
AI lipsync is a gateway to editing / revising any human in any way in any video.
can't wait to show you what we've been working on
made w/ sync-1.6.1
@synclabs_so
⚡
5
9
46
blessed to be part of the story
it’s getting good 🚀
0
1
43
⚡️
@FakeTimeAI
wk3 update:
This week was rough. We had some big wins + some big challenges.
[build] video works – almost uncannily good. voice cloning + transcription are also pretty solid. But still working out a bug connecting our front + backend.
[launch] went viral. twice
5
3
40
how does it work?
[1] make a post request w/ links to a hosted audio + video file
[2] we lip-sync video / image to audio, no training required
[3] get a link to a synchronized video → [future] stream video back
detailed guide here:
2
4
36
you don’t need New Year’s resolutions when you are fully resolved
here’s to 2024 🥂
lets accelerate
2
1
37
this is gonna be sick
dm if you wanna hack w/ us at sync house™ 😈
⚡lipsync-1.5-beta is live
the past few weeks were wild:
– our team moved to SF + setup the hacker house
– kicked off
@ycombinator
w24 batch
– launched an updated lipsync model + subscriptions
– crossed 10k+ users / 160hrs+ generated weekly 🚀
catch up w/ us in sync.log 004 👇
2
5
46
4
0
37
⚡️ explored personalized product marketing w/ deepfake models this weekend
what if brands turned us into our own heroes?
would you want this, or is this going too far? 🤔
7
0
37
⚡️
@FakeTime
Update
Just chatted w/
@KimKardashian
– she loves
@FakeTimeAI
🚀
What if creators could build + scale 1:1 relationships w/ fans + fans could learn & grow from the people they idolize 🤔
cc.
@_buildspace
|
@_nightsweekends
Wanna try it out? Link below👇
6
3
33
visual dubbing is now mainstream
honestly, can't be more excited about this
bad dubs plague every consumer outside the west – the fact this is changing shows how markets are moving
looking forward to a world where every consumer is a first class citizen
1
1
34
just helped snickers clone Josè Mourinho 👌🏼
4
2
33
imagine training models on encrypted data
or generating embeddings + retrieval w/o reading underlying data
massive potential to overcome privacy issues w/ data in AI
math is beautiful.
Why is Fully Homomorphic Encryption (FHE) the next big thing 🧵
25
55
298
2
0
30
some of the hungriest / best founders I know 🚀
Our YC application video.
A bank-built-in WhatsApp for Latam gig workers.
We are transforming the Latin American fragmented system🚀
This’s how it starts…
64
28
566
2
1
27
^this is pt.1 → pt.2 to follow (twitter can't handle massive video files yet)
if you wanna try dubbing + lipsyncing your content yourself try out our new repl
its trending on
@Replit
's home page rn :)
2
2
27
ayo this was awesome
@safwaankay
/
@0xSamHogan
ty for having me 🙏🏼
just thinking / learning out loud
ATTENTION HEADS – EP 9.0
@0xSamHogan
and I sit down with
@therealprady
, cofo + CEO of Sync, a real time lip-syncing API.
We discuss:
- Personalized ads
- Challenges of building a foundational model
- The paper I saw but didn’t read (a model to detect smells)
LINK ->
3
0
27
0
1
26
insanely fun to ship this
first of many iterations we're working on for chapter 2 of sync
also check out the discord bot - too much to play around w/ 😂
lil toy we're building for
@_nightsweekends
+
@_buildspace
⚡today we're launching lipsync-1.7.0-beta
an experimental state-of-the-art video-to-video lipsync model that generates more natural teeth and accurate skin tones
also generates ~1.5x faster than 1.6.0 and ~2x faster than 1.6.1 👀
available now – also through our new discord
2
25
96
2
0
23
PSA: life is a celebration
As builders it’s easy to get lost in our intensity / passion + focus to execute
Life’s bigger than just the things we create — it’s about who we create w/ + build for that matters the most
Remember to take time for moments that matter
Blessed to
0
0
22
going into this eyes wide open w/ slight impostor syndrome + hunger to learn / fucking ship
0
0
22
this is sick 👌🏼
made w/
@synclabs_ai
wonder if you could bring animatediff characters to life too
1
0
22
slow is smooth and smooth is fast → learn what users want + build it
vs
move fast + break things → ship shit + see what sticks
what's better? it depends
2
2
21
⚡️ grateful to launch something personal today
Felt inspired to do a side project + create something for its own sake — not a product, maybe it’s art?
Both to represent me digitally + explore an idea: right vs left 🧠 — the two sides of a creator.
cc.
@_buildspace
/
6
0
19
love this 👌🏼
always awesome to see friends using tools you build
Trying out
@therealprady
and
@noahmilberger
’s awesome lip syncing tech with some audio translation mixed in. So cool!
1
1
11
0
0
20
ps. to the new owners of my sim2s, miuras, bettinardis, and red scotty
post ur next round
hope you can handle them
8
0
19
1
0
20
isn’t it weird that languages diverge > converge?
it’s like entropy affects more than just physical things
eg. ideas, religions, movements, art
it takes incredible effort to reorder the physical + metaphysical
3
2
20
⚡️
@FakeTimeAI
update: sometimes you need to put your head down + just grind it out
this week set up the foundation for growth in every week to come
[build]
- prototype -> production. v1.0 stable build of FakeTime Audio [b2c]
- generated 1hr+ long-format AI-clone podcast
2
2
18
lipsync can be broken into two problems:
[1] lipsync accuracy / [2] visual quality
this new model has very good accuracy – visual quality is decent, but has room to improve
its solid progress. unlocking a lot of interesting ideas / insights along the way
🚀 quick ship: sync-1.5 in beta
just released beta access to an updated lipsync model
~40% better lipsync accuracy + reduced bounding box artifact
try it out 👇
4
1
23
2
0
19
@asimfiles
@Davidjpark96
it was inevitable
the focus needs to be on things beyond wrappers of existing models — that’s why we’re building our own audio / video models
@synclabs_ai
starting w/ lipsyncing
2
1
19
Retention is hard. Key is to solve for a problem, not just virality.
Don't be fooled by these fast growing AI apps.
They are dying even faster.
Here's some perspective on the massive churn & retention problems in AI:
1/8:
18
124
965
0
1
17
@subby_tech
/
@amit_ajwani
— this is what
@paulg
would look like w/ wav2lip 2.0
HD + more accurate lipsync w/ fewer artifacts
let’s plug it in. good thing were neighbors 👀
@_buildspace
5
1
18
clone it yourself here:
1
2
18
disclaimer: this is an early alpha test of our API
we’re giving it away for free for a limited time to learn how you use it, and understand how to make it better.
this is a powerful tool, but tools don’t have a say in how they’re used. be conscientious + use responsibly.
1
0
15
why is this important? imagine a world where language is no longer a barrier
where we watch our favorite movies in our native language
or learn deep learning from MIT in a language we understand best
even zoom anyone in the world, speak our native tongue + them speak theirs
1
0
15