Echo Scribe: Your API Key&Data

wean

About this app

🚀 Open Source & BYOK AI Assistant
EchoScribe summarizes audio, text, URLs, PDFs and shared messages using your own AI API keys or optional Local AI endpoints. No backend, no tracking, no ads, no subscriptions.

GitHub: https://github.com/vibecodingwean/echoscribe
Info: https://app.wean.de

🎙️ Summary & Transcription
Record audio, share WhatsApp voice messages, paste text, or send URLs to EchoScribe. Choose OpenAI, Gemini, xAI Grok, ElevenLabs Scribe v2, or a Local AI Whisper endpoint. Claude is text-only for summaries and translation.

⚡ Realtime Speech
Enable OpenAI Realtime for live speech-to-text with GPT Live Transcribe, or pick a language for live Realtime Translation. Gemini: off is 3.5 Transcribe file STT with speakers, on is 3.5 Transcribe Live. ElevenLabs: off is Scribe v2 file STT, on is Scribe v2 Realtime. Text appears as it arrives.

💻 Desktop & Chrome
The Linux/GNOME companion adds push-to-talk dictation and clipboard paste. Downloads and setup are on GitHub. Web Summary is the Chrome extension.

🧠 Local AI
Use your own Ollama-compatible /api/chat endpoint for summaries and translations, plus an OpenAI-compatible Whisper endpoint for speech-to-text. The Linux installer can recommend models from detected CPU, RAM, GPU and VRAM, and can optionally request extra CanIRun.ai suggestions after showing the hardware values that would be sent. Keep endpoints on a trusted LAN or VPN.

📱 Android Keyboard & Dictation
Enable EchoScribe as a system keyboard (QWERTY English or QWERTZ Deutsch) with an AI toolbar to rewrite, restyle, translate, and reply from the clipboard. Typing stays on-device. EchoScribe Voice Mode can transcribe with OpenAI, Gemini, Grok, Local AI, or ElevenLabs Scribe v2.

An optional floating dictation button works in editable Android fields. Recording starts only after you tap the microphone, a preview is shown, and text is inserted only after you approve it. Android Accessibility is used only to detect editable fields, show the button, and insert approved text. Password, PIN, payment, banking, credit-card and phone fields are blocked.

🔒 Privacy & Security
• No data is sent to EchoScribe servers because there are none.
• AI requests go directly from your device to the provider or Local AI endpoint you select.
• API keys are stored encrypted with Android Keystore.
• You control costs and limits in your provider dashboard.
• You are responsible for your own API keys, endpoints and usage costs.

🎨 AI Image Generation
Create images from prompts in the Bubble Dock. Supports OpenAI, Gemini and Grok image models (Grok Pro: Imagine Image 2.0), with direct sharing and saving.

🔊 Audio, TTS & Summaries
Record audio, transcribe, summarize, translate, copy, share, or listen with supported TTS. Share can send text, generated TTS audio, or a generated image to another app. History stays on device.

🔄 Workflows
• Audio Summary: record, transcribe and summarize in one flow.
• Text Summary: paste or share text from other apps.
• URL Summary: extract webpage content locally before sending it to your provider.
• Translation: change the target language and re-process the original source.
• Custom Prompts: adjust summary, URL, audio and dictation prompts.

🤖 Providers & Models
Fast presets: GPT-5.6 Terra, Gemini 3.8 Flash, Claude Sonnet 5 and Grok 4.3.
Pro presets: GPT-6 Astra, Gemini 3.1 Pro Preview, Claude Opus 5 and Grok 4.6. Speech models include GPT Transcribe, GPT Live Transcribe, ElevenLabs Scribe v2 and optional Scribe v2 Realtime.

🔑 API Key Setup
• OpenAI: platform.openai.com
• Gemini: aistudio.google.com
• Claude: console.anthropic.com
• Grok: console.x.ai
• ElevenLabs STT + TTS: elevenlabs.io/app/settings/api-keys
• Local AI: configure your own Ollama and Whisper URLs

✨ Simple & Transparent
One-time purchase. No subscriptions, no ads, no tracking. API costs are billed directly by your selected AI provider.

What’s new

OpenAI Pro is GPT-6 Astra for summary and translation. Fast stays GPT-5.6 Terra.