Add AI Pronunciation Diagnostics to Any Application
Empower your users with real-time, phoneme-level speech breakdown. Seamlessly integrate our fast diagnostic API into your EdTech platform or language app.
// Request Payload (FormData with WAV audio)
const result = await client.diagnose({
audio: audioBuffer,
referenceText: "The weather is beautiful.",
language: "en-US"
});
// Real-time Phonetic Breakdown Output
{
"status": "success",
"overall_score": 92.4,
"phonetic_accuracy": 94.1,
"words": [
{
"word": "beautiful", "ipa": "/bjuːtɪfʊl/", "score": 62,
"phonemes": ["b", "j", "uː", "t", "ɪ", "f", "ʊ", "l"]
}
]
}
Granular Analysis
From Sentence Fluency to Phoneme Probability
Our models do not just return a simple pass or fail grade. They mathematically dissect speech across multiple dimensions to pinpoint the exact acoustic root of a mispronunciation.
Sentence Level Profiling
Evaluates overall prosody, intonation contours, and speech rate (WPM & syllables per second) against native baseline averages.
Word Level Isolation
Scores individual word clarity and identifies misplaced lexical stress markers or missing syllable omissions.
Phoneme Probability Matrix
Generates statistical confidence percentages for expected phonemes versus erroneously substituted phonemes using acoustic formant (F1/F2) mapping.
Engine Capabilities
Deep Speech Diagnostics, Simple Integration
We handle the complex machine learning, signal processing, and acoustic modeling so you can deliver personalized feedback to your learners.
Phoneme-Level Breakdown
Identify exact mispronunciations, vowel substitutions, and omitted consonants using standard International Phonetic Alphabet (IPA) representation.
Ultra-Low Latency Edge API
Global edge deployment returns diagnostic results in under 280ms, enabling real-time voice feedback inside mobile apps and web platforms.
Multi-Accent & Dialect Model
Trained on over 60,000 hours of non-native speech across 40+ global languages to ensure equitable scoring for non-native learners.
Privacy & Zero-Retention Option
Full GDPR & COPPA compliance. Voice payloads are evaluated strictly in memory and purged immediately unless explicitly saved by your application.
Simple REST & WebSockets
Send audio in standard WAV, MP3, WebM, or OGG containers. Receive lightweight JSON payloads formatted specifically for EdTech dashboards.
Fluency & Pitch Prosody
Measure speech rate (syllables per second), pause durations, intonation contours, and stress placement for comprehensive fluency diagnostics.
Built for developers, designed for speed
Integrate the diagnostic engine into your stack with just a few lines of code. SDKs available for JavaScript, Python, Swift, and Kotlin.
Capture Audio Stream
Record standard PCM/WAV audio via standard HTML5 or native mic APIs.
Post to Diagnostic Endpoint
Pass the audio payload and target reference string to our API.
Render Highlighted Feedback
Map word scores and phoneme tags directly to your user interface.
import { AdvancedLearningClient } from '@advancedlearning/sdk';
const client = new AdvancedLearningClient({
apiKey: process.env.DIAGNOSTIC_API_KEY
});
async function evaluateSpeech(audioBlob, targetText) {
const response = await client.diagnose({
audio: audioBlob,
referenceText: targetText,
language: 'en-US',
phonemeGranularity: 'detailed'
});
console.log(`Overall Accuracy: ${response.overall_score}%`);
return response.words;
}import requests
api_url = "https://api.advancedlearning.tech/v1/analyze"
headers = {"Authorization": "Bearer YOUR_API_KEY"}
files = {'audio': open('user_speech.wav', 'rb')}
data = {
'reference_text': 'The quick brown fox',
'language': 'en-US'
}
response = requests.post(api_url, headers=headers, files=files, data=data)
result = response.json()
print(f"Score: {result['overall_score']}")curl -X POST https://api.advancedlearning.tech/v1/analyze \
-H "Authorization: Bearer YOUR_API_KEY" \
-F "audio=@user_recording.wav" \
-F "reference_text=The quick brown fox" \
-F "language=en-US"Simple & Predictable
Pay Only For What You Diagnose
Disruptive EdTech pricing starting at $0.09 per 100 API calls. No monthly software tier lockups.
Interactive Volume Cost Calculator
Drag the slider to calculate estimated monthly diagnostic costs.
Standard API
Ideal for growing language platforms, startups, and mobile applications.
- Full phoneme & word level IPA feedback
- All 40+ supported language models
- Up to 30 seconds of audio per call
- Standard Community & Discord Support
Frequently Asked Questions
Technical details regarding accuracy, security, and payload structures.