Andreas Trolle

Just released a video on how to build a cross-platform mobile app for learning Mandarin. ↓ Link below ↓ 🎧 Listening + speaking practice modes with progress tracking 🎙️ Voice recording and AI transcription for pronunciation practice 💬 Real-time AI roleplay conversations with scenario goals/tasks ✨ Custom scenario generation 📈 Speaking/listening stats and lesson completion tracking 📱 Cross-platform iOS/Android app with Expo 🔐 Passwordless authentication with Supabase magic links 🧭 Personalized onboarding flow (level, motivation, interests) 📚 Structured Mandarin curriculum (12 chapters, 86 lessons) 💳 In-app paywall flow

New tutorial is out! Build a HeyGen clone, that generates realistic AI avatar videos. ↓ Link below ↓ 🧑‍🦰 AI Portrait Avatars with fudan-generative-ai/hallo3 🗣️ AI Voice Cloning & TTS with chatterbox-tts 🌐 AI Video Translation with lip-sync 🔄 Change Video Audio with automatic lip-syncing ⚡ Serverless GPU Processing with @modal 💽 Postgres database with @neondatabase 📊 Background Job Queue 💳 Credit-Based System for video generation 💰 Polar.sh Integration for purchasing credit packs 👤 User Authentication with @better_auth 🎛️ Dashboard to manage generated videos 🐍 Python & FastAPI Backend for AI logic 📱 Modern UI with Next.js, Tailwind CSS & Shadcn UI

@miketennys76930 Thank you for that! I just moved to another country so upload frequency will be less for the rest of this year. Next one going up in 2 weeks!

@Andreaswt21 @Andreaswt21 When will you put a new video T_T .. have been following all your videos and i know you publish each month.. but its been over a month this time.. every day i refresh your page in hope to see your new upload x.x .. Can you let us know when will you be uplaoding?

My new tutorial is out - learn to build a complete AI music generation SaaS that generates original music using text prompts. ↓ Link below ↓ 🎵 AI Music Generation with ACE-Step 🧠 LLM-Powered Lyric & Prompt Generation with Qwen2-7B 🖼️ AI Thumbnail Generation with stabilityai/sdxl-turbo 🎤 Multiple Generation Modes for descriptions, custom lyrics, or described lyrics 🎸 Instrumental Tracks option to generate music without vocals ⚡ Serverless GPU Processing with @modal_labs for fast generation 📊 Queue System with @inngest for handling generation tasks in the background 💳 Credit-Based System 💰 @polar_sh Integration for purchasing credit packs 👤 User Authentication with @better_auth 💽 Database by @neondatabase 🎧 Community Music Feed to discover, play, and like user-generated music 🎛️ Personal Track Dashboard to manage, play, and publish songs 🐍 Python & FastAPI Backend for music generation logic 📱 Modern UI with Next.js, Tailwind CSS & Shadcn UI

@Elnora_Hines18 Great job!

Finished up with training of the largest model I have built till now. It's an audio classifier based on CNN . Thanks @Andreaswt21 for the tutorial. I learned a lot from it.

@b_nandhan68170 Awesome!

Massive shoutout to @Andreaswt21 His architecture walkthrough made this possible. Built this to learn & level up—hope it helps others too! #AI #LLM #whisperX #Gemini #OpenCV #FastAPI #T3Stack #Modal #Stripe #SaaS #DevBuild

🚀 Built an AI-Powered Podcast Clipper — transforms long podcasts into short, viral clips for TikTok, Shorts & Reels. It uses AI + LLM + CV to detect viral moments, track speakers & crop clips. Here’s how it works under the hood 👇

I just released a video where I train a Convolutional Neural Network from scratch and build a dashboard to visualize its layer outputs. Including 🧠 Deep Audio CNN for sound classification 🧱 ResNet-style architecture with residual blocks 🎼 Mel Spectrogram audio-to-image conversion 🎛️ Data augmentation with Mixup & Time/Frequency Masking ⚡ Serverless GPU inference with @modal_labs 📊 Interactive Next.js & React dashboard 👁️ Visualization of internal CNN feature maps 📈 Real-time audio classification with confidence scores 🌊 Waveform and Spectrogram visualization 🚀 FastAPI inference endpoint ⚙️ Optimized training with AdamW & OneCycleLR scheduler 📈 TensorBoard integration for training analysis 🛡️ Batch Normalization for stable & fast training 🎨 Modern UI with Tailwind CSS & Shadcn UI ✅ Pydantic data validation for robust API requests

@mohammadabbas42 Appreciate it!

@Andreaswt21 Love what you do brother.

My latest tutorial is out, on building a SaaS app that converts podcasts into viral short-form clips, ready for YouTube Shorts and TikTok. ↓ Link to video below ↓ 🎬 Auto-detection of viral moments in podcasts (stories, questions, etc.) 🔊 Automatically added subtitles on clips 📝 Transcription with m-bain/whisperX 🎯 Active speaker detection for video cropping with Junhua-Liao/LR-ASD 📱 Clips optimized for vertical platforms (TikTok, YouTube Shorts) 🎞️ GPU-accelerated video rendering with FFMPEGCV 🧠 LLM-powered viral moment identification with Gemini API 📊 Queue system with @inngest for handling user load 💳 Credit-based system 💰 @stripe integration for credit pack purchases 👤 User authentication system 📱 Responsive @nextjs web interface 🎛️ Dashboard to upload podcasts and see clips ⏱️ Inngest for handling long-running bg processes ⚡ Serverless GPU processing with @modal_labs 🌐 @FastAPI endpoint for podcast processing 🎨 Modern UI with @tailwindcss & @shadcn

@modal_labs 👏

Don't sleep on @Andreaswt21 's channel. His latest video is a detailed walkthrough of a biotech web app using Evo 2 to do variant effect prediction. 🧬 We're super impressed with Andreas's hustle and are proud to supply the underlying compute 📦

TL;DR DNA is like a long code made of A, T, G, and C's. Small changes (mutations) in specific parts of this code, like in genes responsible for preventing cancer, can increase a person's risk of developing the disease. For instance, if an 'A' appears where a 'T' should be at a particular spot, that's a mutation. We'll build a tool to predict whether a mutation is likely to cause diseases or not.

I recently released a tutorial on building a biotech web app that can classify how likely specific mutations in DNA are to cause diseases (variant effect prediction). No AI or Biotech background is needed, since I’ll cover all the theory needed. The pathogenicity prediction is done with Evo 2 - a recently released DNA language model from @arcinstitute ↓ Link to video below ↓ 🧬 Evo 2 model for variant effect prediction 🩺 Predict pathogenicity (pathogenic/benign) of single nucleotide variants (SNVs) ⚖️ Comparison between existing ClinVar classification and Evo 2 prediction 💯 Prediction confidence estimation 🌍 Genome assembly selector (e.g., hg38) 🗺️ Select genes from chromosome browsing or searching (e.g., BRCA1) 🌐 See full reference genome sequence (UCSC API) 🧬 Explore gene and variants data (NCBI ClinVar/E-utilities) ⚡ Deploy AI model on an H100 serverless GPU for FREE with @modal_labs 🚀 @FastAPI endpoint for variant analysis requests 📱 @nextjs web app based on create-t3-app from @theo 🎨 Modern UI with @tailwindcss & @shadcn UI