{"slug": "create-a-meditation-app-with-ai-generated-voice", "title": "Create a Meditation App with AI-Generated Voice", "summary": "A developer published a walkthrough for building a meditation app that uses ElevenLabs' text-to-speech API to generate natural-sounding guided narration. The tutorial covers cloning a voice from a short audio sample to obtain a voice_id, then calling the synthesize endpoint with stability and similarity_boost settings to produce MP3 audio for guided breathing scripts. It includes Python and curl examples for both the voice-cloning and synthesis steps.", "body_md": "If you’ve ever built a simple chatbot or a notification system, you’ve already worked with APIs, authentication, and the occasional rate limit. Adding a *real‑time, natural‑sounding voice* turns a text‑based experience into something that feels comforting, personal, and almost therapeutic. For meditation, that’s a game‑changer: a gentle narrator can guide breathing, set intentions, or play ambient sounds—all while keeping your users engaged.\n\nIn this post we’ll walk through:\n\nBy the end, you’ll have a working prototype that you can extend into a full‑featured meditation app.\n\nText‑to‑speech (TTS) is a mature field, but most free or open‑source solutions lag behind commercial providers in naturalness, voice variety, and developer experience. ElevenLabs offers:\n\nIf you’re looking for a quick, production‑ready voice solution, check out ElevenLabs at [https://try.elevenlabs.io/kr07zfuqn1bp](https://try.elevenlabs.io/kr07zfuqn1bp). The free tier is generous, and you can upgrade to a paid plan for higher quality and more requests.\n\nA generic “calm” voice can work, but a cloned voice that mimics the user’s own voice or a brand‑specific narrator creates a deeper connection. ElevenLabs’ cloning workflow is simple:\n\n`voice_id`.\nBelow is a quick Python script to upload a clip and get the `voice_id`:\n\n``` python\nimport os\nimport requests\n\nAPI_KEY = os.getenv(\"ELEVENLABS_API_KEY\")\nUPLOAD_URL = \"https://api.elevenlabs.io/v1/voices\"\n\nheaders = {\"xi-api-key\": API_KEY, \"Accept\": \"application/json\"}\n\n# 1️⃣ Record a short sample (e.g., 30s) and save as `sample.wav`\nwith open(\"sample.wav\", \"rb\") as audio:\n files = {\"file\": (\"sample.wav\", audio, \"audio/wav\")}\n response = requests.post(f\"{UPLOAD_URL}/clone\", headers=headers, files=files)\n\nif response.ok:\n voice_id = response.json()[\"voice_id\"]\n print(\"Your cloned voice ID:\", voice_id)\nelse:\n print(\"Error:\", response.text)\n```\n\nRemember to keep the audio file under 3 MB for the free tier. If you need larger samples, upgrade your plan.\n\nOnce you have a `voice_id`, you can use it in subsequent synthesis calls to produce a voice that *sounds* like the original speaker.\n\nA typical meditation routine might consist of:\n\nWe’ll generate these on demand using ElevenLabs’ `synthesize` endpoint.\n\n``` python\nimport os\nimport requests\n\nAPI_KEY = os.getenv(\"ELEVENLABS_API_KEY\")\nVOICE_ID = os.getenv(\"VOICE_ID\") # The cloned voice ID you got earlier\nTTS_URL = f\"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}\"\n\nheaders = {\n \"xi-api-key\": API_KEY,\n \"Content-Type\": \"application/json\"\n}\n\ndef synthesize(text, output_path=\"output.mp3\"):\n payload = {\n \"text\": text,\n \"voice_settings\": {\n \"stability\": 0.5, # 0-1.0\n \"similarity_boost\": 0.75\n }\n }\n response = requests.post(TTS_URL, headers=headers, json=payload, stream=True)\n if response.status_code == 200:\n with open(output_path, \"wb\") as f:\n for chunk in response.iter_content(chunk_size=8192):\n f.write(chunk)\n print(f\"Saved audio to {output_path}\")\n else:\n print(\"Synthesis failed:\", response.text)\n\n# Demo: a short guided breathing session\nscript = \"\"\"\nWelcome. Let's take a moment to settle in. \nClose your eyes, and breathe in slowly through your nose.\nHold for a count of three.\nNow exhale gently through your mouth.\nRepeat this cycle for a few minutes.\n\"\"\"\n\nsynthesize(script)\n```\n\nThis script pulls the voice model from ElevenLabs and writes the resulting MP3 to disk. You can adapt the `synthesize` function to stream audio directly to a mobile app or a web player.\n\n`curl` for Quick Tests\nIf you prefer the CLI:\n\n```\ncurl -X POST \"https://api.elevenlabs.io/v1/text-to-speech/$VOICE_ID\" \\\n -H \"xi-api-key: $ELEVENLABS_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"text\": \"Hello from ElevenLabs! This is a test of the TTS engine.\",\n \"voice_settings\": { \"stability\": 0.5, \"similarity_boost\": 0.75 }\n }' --output test.mp3\n```\n\nLet’s assume you’re building a React Native app. The simplest way to play synthesized audio is to:\n\n`expo-av` or `react-native-sound`.\nHere’s a quick React Native snippet:\n\n``` python\nimport React, { useEffect } from 'react';\nimport { View, Button } from 'react-native';\nimport { Audio } from 'expo-av';\n\nexport default function MeditationScreen() {\n const playAudio = async () => {\n const { sound } = await Audio.Sound.createAsync(\n { uri: 'https://your-backend.com/audio/meditation.mp3' },\n { shouldPlay: true }\n );\n // Optionally unload when finished\n sound.setOnPlaybackStatusUpdate((status) => {\n if (status.didJustFinish) {\n sound.unloadAsync();\n }\n });\n };\n\n return (\n \n