Moshi AI vs Async (formerly Podcastle)

Moshi AI is a speech-native conversational model from Kyutai, a Paris-based open-science research lab. Instead of chaining speech recognition, text generation, and text-to-speech, Moshi processes audio directly and holds full-duplex voice conversations with minimal latency.

Its multi-stream design runs separate channels for the user, Moshi's spoken output, and an Inner Monologue text stream that improves coherence. That setup lets Moshi listen and talk at the same time, handle overlaps, interruptions, and backchanneling like a real conversation rather than rigid speaker turns.

Moshi is built on Helium, a 7B language model, and Mimi, Kyutai's neural audio codec. Weights and inference code ship for PyTorch, Rust, and MLX, and you can try it in the browser at moshi-chat.kyutai.org. Researchers, voice AI developers, and anyone building real-time spoken interfaces will find the most value here.

Async lets you record, edit, and republish audio and video by chatting with AI instead of dragging clips on a timeline. You can capture remote podcasts, generate footage from prompts, clean up audio, add subtitles, dub content, and turn long recordings into social clips inside one browser workflow.

Where most editors split recording, transcription, and clip creation across separate apps, Async bundles them behind a chat interface called Bumblebee. You upload footage or start from a template, tell Async what to cut or generate, and it handles pacing, captions, and export. The platform also lists 100+ generative models for video, image, music, and voice, including Kling, Veo, Sora, and GPT Image 2.

The free plan includes 10 lifetime AI credits and 1 hour of media uploads, enough to test the chat editor before upgrading. Paid tiers add monthly credit pools (450 on Essentials, 1,200 on Pro, 3,000 on Teams) that meter video generation, dubbing, and other AI actions. Annual billing saves up to 40%.

Podcasters, video creators, marketing teams, and social media managers who want recording, editing, and repurposing in one place without learning traditional NLE software are the target audience. An iOS app is available, and remote recording supports up to 10 participants per session.