--- title: "AI Voice Generator and Text to Speech" canonical: https://withicarus.com/audio site: Icarus description: "Text to speech in natural voices for ads, reels and product videos. Voices from Google, ElevenLabs, MiniMax, Qwen and ByteDance." --- # AI Voice Generator and Text to Speech Text to speech in natural voices for ads, reels and product videos. Voices from Google, ElevenLabs, MiniMax, Qwen and ByteDance. Type a script, pick one of 506 voices, and Icarus reads it out loud. Play it, then download it for your ad, reel or product video. Try it: https://withicarus.com/dashboard/audio?tab=speech ## How it works 1. Type your script, or paste it in. 2. Pick a voice, and the AI that reads it if you want a different one. 3. Press Generate. Play it, then download it. ## Which to pick - **Gemini 3.8 Flash Speech** (Google): Two-voice conversations, steered by a written note. 30 voices, scripts to 5,000 characters, conversations of 2 voices, $0.03 for 1,000 characters. - **Eleven v4** (ElevenLabs): Natural, expressive reads you steer with tags like [whispers]. 42 voices, 84 languages, scripts to 5,000 characters, $0.10 for 1,000 characters. - **Eleven v4 Turbo** (ElevenLabs): The same natural reads, back sooner and at half the price. 42 voices, 84 languages, scripts to 5,000 characters, $0.05 for 1,000 characters. - **Eleven v3** (ElevenLabs): Conversations between up to 10 voices, with tags like [laughs]. 42 voices, 85 languages, scripts to 5,000 characters, conversations of 10 voices, $0.13 for 1,000 characters. - **MiniMax Speech 2.8 HD** (MiniMax): Long scripts, with the mood and tone set by you. 348 voices, 37 languages, scripts to 10,000 characters, $0.13 for 1,000 characters. - **MiniMax Speech 2.8 Turbo** (MiniMax): The same voices and controls, back sooner. 348 voices, 37 languages, scripts to 10,000 characters, $0.08 for 1,000 characters. - **Qwen Audio 3** (Qwen): Short, bright lines in 10 languages. 45 voices, 10 languages, scripts to 2,000 characters, $0.07 for 1,000 characters. - **Seed Speech 2** (ByteDance): The lowest price, steered by a written note. 41 voices, 10 languages, scripts to 5,000 characters, $0.04 for 1,000 characters. - **Seed Audio 1** (ByteDance): Speech shaped by a written scene. 21 voices, scripts to 2,048 characters, $0.004 for a second. Every price: https://withicarus.com/pricing#audio-costs-heading ## Use it from Claude or ChatGPT Connect Icarus once (https://withicarus.com/mcp), then ask in plain words. The file comes back in the chat. ## More audio - [AI Voice Changer for Recordings and Videos](https://withicarus.com/audio/voice-changer): Swap the voice in a recording or a video. Your words and timing stay. - [AI Video Translator: Translate and Dub Video](https://withicarus.com/audio/video-translator): Your video, speaking another language. ## Questions ### What is an AI voice generator? A tool that reads your words out loud in a natural voice. You type the script, pick the voice, and download the recording. ### How many voices and languages can I pick from? 506 voices from Google, ElevenLabs, MiniMax, Qwen and ByteDance. Eleven v3 speaks the most languages: 85. ### Can two people talk in one clip? Yes, on Gemini 3.8 Flash Speech (2 voices) and Eleven v3 (10 voices). Give each line its own voice. ### How long can my script be? 10,000 characters on MiniMax Speech 2.8 HD and MiniMax Speech 2.8 Turbo. Each AI's own limit is listed on this page. ### What does it cost? Each AI has its own price, listed on this page and on the pricing page. You see the price before you press Generate. ### Can I use it from Claude or ChatGPT? Yes. Connect Icarus to Claude or ChatGPT once, then ask for a voiceover in plain words. The file comes back in the chat. --- Site: Icarus (https://withicarus.com) Canonical: https://withicarus.com/audio Contact: tanmay@hooked.design