
BREAKING: Big day for voice AI. @Cartesia just dropped Sonic 3.5 and @SarvamAI announced their $300M Series B - so I decided to put both to the test myself. Same scripts, side by side. The results…
The Hook
“नमस्ते India! मैं हूँ आपकी नई voice assistant, और मैं सिर्फ़ Hindi नहीं, आपकी हर भाषा में बात कर सकती हूँ, उसी अपनेपन के साथ। आप Delhi से हों, Chennai से, या Kolkata से, मैं आपको वैसे ही समझूँगी जैसे कोई अपना। तो चलिए, शुरू करते हैं।”
Product Reveal
“नमस्ते India! मैं हूँ आपकी नई voice assistant, और मैं सिर्फ़ Hindi नहीं, आपकी हर भाषा में बात कर सकती हूँ, उसी अपनेपन के साथ। आप Delhi से हों, Chennai से, या Calcutta से, मैं आपको वैसे ही समझूँगी जैसे कोई अपना। तो चलिए, शुरू करते हैं।”
Feature Teaser
“(No spoken dialogue — The driving electronic beat with sub-bass and arpeggiated synths continues, maintaining an energetic and futuristic atmosphere.)”
Synthesizable in Remotion · key components of the split-screen-ai-comparison template.
A vertical split screen divides the frame, with each side featuring a circular audio visualizer whose bars dynamically extend and retract based on audio input. Text overlays identify each product.
Utilize Remotion's <AbsoluteFill> for the split. Implement two <AudioVisualizer> components, each with a circular SVG mask and bars animated using `interpolate` based on `useCurrentFrame().getAudioData()`. Text elements are positioned with `absolute` styling.
A circular button with a play or pause icon appears, changing state to reflect the audio playback status. It has a subtle glow or pulse effect.
Create a circular SVG or div with a background color. Use conditional rendering to switch between a play triangle SVG and a pause double-bar SVG. Apply a `spring` animation for scale and opacity on state change, and a `glow` CSS filter for the pulse.
A small, colored dot precedes product names, providing a clean, minimalist visual identifier. It remains static throughout the scene.
Implement a simple `div` with `borderRadius: '50%'` and a fixed `backgroundColor`. Position it using `flexbox` or `absolute` positioning next to the product name text. No animation is applied to this primitive.
Purpose. Introduce the two competing AI products and the premise of the comparison.
Execution. Split-screen with product names and initial static UI elements.
Purpose. Showcase the first AI's voice capabilities and responsiveness.
Execution. Left screen's visualizer activates, displaying the 'robotic' voice.
Purpose. Showcase the second AI's voice capabilities, highlighting its naturalness.
Execution. Right screen's visualizer activates, displaying the 'natural' voice, while the left pauses.
Purpose. Allow the viewer to directly compare the two AI performances.
Execution. Both visualizers continue to react, with the right side showing more dynamic activity, reinforcing the 'natural' voice.
Objective
A new language learning app called 'LinguaFlow' wants to compare its 'Immersion Mode' with a 'Traditional Grammar' approach.How the recipe adapts
The split-screen layout will compare 'Traditional Grammar' on the left with 'Immersion Mode' on the right. Instead of audio visualizers, the left will feature a progress bar filling up, and the right will have a dynamic word cloud forming from spoken words. The product identifier dots will be replaced by small icons representing each learning method. The voiceover will explain the benefits of each method, followed by a visual demonstration of their respective UI elements in action, all set against the energetic electronic beat.Scene durations (3 scenes, 35.9s)
Key Remotion components
To ensure that artificial general intelligence benefits all of humanity

Gopuff introduces Go, an AI shopping assistant built with SpaceXAI: say what you need and the order is placed.

Iliad (Troy) trailer made by Grok Imagine 1.5, which was just released