Speech-to-Text

STT HTTP Examples

Copy page

Complete, copy-paste ready examples for transcribing audio files using HTTP.

Quick Start

Basic File Transcription

Code
 
curl https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://cdn.openai.com/whisper/draft-20220913a/micro-machines.wav","language":"en"}'

Response:

Code
 
{{
  "text": "Micro-Machine Man, presenting the most midget...",
  "language": "en",
  "segments": [
    {
      "start": 1.09,
      "end": 30.302,
      "text":"Micro-Machine Man, presenting the most midget...",
      "speaker": null
    },
    {
      "start": 30.594,
      "end": 58.558,
      "text": " Where is the boat with the vehicle, and the man with the gunpowder at the end of his screen? You call it the car wash. Where is the toll bridge? And these play sets fit together to form a micromachine world. Micromachine pocket play sets. Such a menace in driving so perfectly precise and dazzlingly detailed. You ought to pocket them all. Micromachines are micromachine pocket play sets. They certainly find a home. The smaller they are, the better they are.",
      "speaker": null
    }
  ],
  "model_time": 2.6508004665374756,
  "audio_duration": 59.77675,
  "duration": 57.467999999999996
}

---

HTTP Examples by Language

cURL

Basic Audio File

Code
 
curl https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"

With Language Hint

Code
 
curl https://api.slng.ai/v1/stt/deepgram/nova:2 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  -F "language=en"

Multiple Audio Formats

Supported formats: MP3, WAV, FLAC, OGG, M4A, WebM

Code
 
# WAV file
curl https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"

# FLAC file
curl https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"

JavaScript/TypeScript

Basic File Upload

Code
 
async function transcribeAudio(audioFile) {
  const formData = new FormData();
  formData.append("file", audioFile);

  const response = await fetch(
    "https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3",
    {
      method: "POST",
      headers: {
        Authorization: "Bearer YOUR_API_KEY",
      },
      body: formData,
    },
  );

  const result = await response.json();
  return result.text;
}

// Usage with file input
const fileInput = document.querySelector('input[type="file"]');
fileInput.addEventListener("change", async (e) => {
  const file = e.target.files[0];
  const transcription = await transcribeAudio(file);
  console.log("Transcription:", transcription);
});

Complete Implementation with Error Handling

Code
 
interface TranscriptionSegment {
  start: number;
  end: number;
  text: string;
}

interface TranscriptionResult {
  text: string;
  segments: TranscriptionSegment[];
  language?: string;
  duration?: number;
}

async function transcribeFile(
  file: File,
  provider: string = "slng",
  model: string = "whisper",
  variant: string = "large-v3",
  apiKey: string = "YOUR_API_KEY",
): Promise<TranscriptionResult> {
  const url = `https://api.slng.ai/v1/stt/${provider}/${model}:${variant}`;

  const formData = new FormData();
  formData.append("file", file);

  try {
    const response = await fetch(url, {
      method: "POST",
      headers: {
        Authorization: `Bearer ${apiKey}`,
      },
      body: formData,
    });

    if (!response.ok) {
      const error = await response.text();
      throw new Error(`HTTP ${response.status}: ${error}`);
    }

    const result = await response.json();
    return result;
  } catch (error) {
    console.error("Transcription error:", error);
    throw error;
  }
}

// Usage
const file = document.querySelector('input[type="file"]').files[0];
const result = await transcribeFile(file);

console.log("Full text:", result.text);
console.log("Segments with timestamps:", result.segments);

Batch Processing Multiple Files

Code
 
async function transcribeBatch(audioFiles) {
  const results = await Promise.all(
    audioFiles.map(async (file) => {
      try {
        const formData = new FormData();
        formData.append("file", file);

        const response = await fetch(
          "https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3",
          {
            method: "POST",
            headers: {
              Authorization: "Bearer YOUR_API_KEY",
            },
            body: formData,
          },
        );

        const result = await response.json();
        return {
          filename: file.name,
          transcription: result.text,
          segments: result.segments,
          success: true,
        };
      } catch (error) {
        return {
          filename: file.name,
          error: error.message,
          success: false,
        };
      }
    }),
  );

  return results;
}

// Usage
const files = Array.from(document.querySelector('input[type="file"]').files);
const results = await transcribeBatch(files);

results.forEach((result) => {
  if (result.success) {
    console.log(`${result.filename}: ${result.transcription}`);
  } else {
    console.error(`${result.filename} failed: ${result.error}`);
  }
});

Using Deepgram Nova for Real-time Files

Code
 
async function transcribeWithDeepgram(audioFile) {
  const formData = new FormData();
  formData.append("file", audioFile);
  formData.append("language", "en");
  formData.append("punctuate", "true");
  formData.append("diarize", "true"); // Speaker identification

  const response = await fetch("https://api.slng.ai/v1/stt/deepgram/nova:2", {
    method: "POST",
    headers: {
      Authorization: "Bearer YOUR_API_KEY",
    },
    body: formData,
  });

  const result = await response.json();
  return result;
}

Python

Basic File Transcription

Code
 
import requests

def transcribe_audio(file_path: str, api_key: str = "YOUR_API_KEY") -> str:
    """
    Transcribe an audio file using SLNG API.

    Args:
        file_path: Path to audio file
        api_key: SLNG API key

    Returns:
        Transcribed text
    """
    url = "https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3"

    headers = {
        "Authorization": f"Bearer {api_key}"
    }

    with open(file_path, "rb") as audio_file:
        files = {"file": audio_file}
        response = requests.post(url, headers=headers, files=files)

    response.raise_for_status()
    result = response.json()

    return result["text"]

# Usage
transcription = transcribe_audio("recording.mp3")
print(transcription)

Complete Implementation with Segments

Code
 
import requests
from typing import List, Dict, Optional
from pathlib import Path

class TranscriptionSegment:
    def __init__(self, start: float, end: float, text: str):
        self.start = start
        self.end = end
        self.text = text

    def __repr__(self):
        return f"[{self.start:.2f}s - {self.end:.2f}s]: {self.text}"

class TranscriptionResult:
    def __init__(self, text: str, segments: List[Dict]):
        self.text = text
        self.segments = [
            TranscriptionSegment(s["start"], s["end"], s["text"])
            for s in segments
        ]

    def get_text_at_time(self, time: float) -> Optional[str]:
        """Get the text being spoken at a specific time."""
        for segment in self.segments:
            if segment.start <= time <= segment.end:
                return segment.text
        return None

def transcribe_file(
    file_path: str,
    provider: str = "slng",
    model: str = "whisper",
    variant: str = "large-v3",
    api_key: str = "YOUR_API_KEY"
) -> TranscriptionResult:
    """
    Transcribe an audio file with full segment information.

    Args:
        file_path: Path to audio file
        provider: STT provider (slng, deepgram)
        model: Model name (whisper, nova)
        variant: Model variant (large-v3, 2)
        api_key: SLNG API key

    Returns:
        TranscriptionResult with text and segments
    """
    url = f"https://api.slng.ai/v1/stt/{provider}/{model}:{variant}"

    headers = {
        "Authorization": f"Bearer {api_key}"
    }

    file_path = Path(file_path)
    if not file_path.exists():
        raise FileNotFoundError(f"Audio file not found: {file_path}")

    with open(file_path, "rb") as audio_file:
        files = {"file": (file_path.name, audio_file, "audio/mpeg")}

        try:
            response = requests.post(url, headers=headers, files=files, timeout=120)
            response.raise_for_status()
            result = response.json()

            return TranscriptionResult(
                text=result["text"],
                segments=result.get("segments", [])
            )

        except requests.exceptions.HTTPError as e:
            print(f"HTTP Error: {e}")
            print(f"Response: {response.text}")
            raise
        except requests.exceptions.RequestException as e:
            print(f"Request Error: {e}")
            raise

# Usage
result = transcribe_file("meeting.mp3")

print("Full transcription:")
print(result.text)
print("\nSegments with timestamps:")
for segment in result.segments:
    print(segment)

# Get text at specific time
text_at_30s = result.get_text_at_time(30.0)
print(f"\nAt 30 seconds: {text_at_30s}")

Batch Processing Directory

Code
 
import os
from pathlib import Path
from concurrent.futures import ThreadPoolExecutor, as_completed

def transcribe_directory(
    directory: str,
    api_key: str = "YOUR_API_KEY",
    max_workers: int = 5
) -> Dict[str, str]:
    """
    Transcribe all audio files in a directory.

    Args:
        directory: Path to directory containing audio files
        api_key: SLNG API key
        max_workers: Maximum concurrent transcriptions

    Returns:
        Dictionary mapping filenames to transcriptions
    """
    audio_extensions = {'.mp3', '.wav', '.flac', '.m4a', '.ogg'}
    directory = Path(directory)

    audio_files = [
        f for f in directory.iterdir()
        if f.suffix.lower() in audio_extensions
    ]

    print(f"Found {len(audio_files)} audio files")

    results = {}

    with ThreadPoolExecutor(max_workers=max_workers) as executor:
        # Submit all transcription tasks
        future_to_file = {
            executor.submit(transcribe_file, str(f), api_key=api_key): f
            for f in audio_files
        }

        # Process completed tasks
        for future in as_completed(future_to_file):
            file_path = future_to_file[future]
            try:
                result = future.result()
                results[file_path.name] = result.text
                print(f"✓ {file_path.name}")
            except Exception as e:
                print(f"✗ {file_path.name}: {e}")
                results[file_path.name] = f"ERROR: {e}"

    return results

# Usage
results = transcribe_directory("./audio_files", max_workers=3)

# Save to file
with open("transcriptions.txt", "w") as f:
    for filename, transcription in results.items():
        f.write(f"=== {filename} ===\n")
        f.write(f"{transcription}\n\n")

print("Transcriptions saved to transcriptions.txt")

---

Model Comparison

SLNG Whisper (High Accuracy)

Best for: Batch transcription, meetings, podcasts, content creation

Code
 
curl https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"

Features:

• Highest accuracy
• Multi-language support (100+ languages)
• Robust to background noise
• Batch processing optimized
• Segment timestamps included

Deepgram Nova (Real-time Optimized)

Best for: Near-real-time file processing, customer calls, IVR recordings

Code
 
curl https://api.slng.ai/v1/stt/deepgram/nova:2 \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  -F "language=en" \
  -F "punctuate=true" \
  -F "diarize=true"

Features:

• Fast processing
• Speaker diarization
• Automatic punctuation
• Word-level timestamps
• Confidence scores

---

Common Parameters

SLNG Whisper

Parameter	Type	Description
file	file	Audio file (required)
language	string	Language code (optional, auto-detected)
temperature	number	Sampling temperature 0-1 (default: 0)

Deepgram Nova

Parameter	Type	Description
file	file	Audio file (required)
language	string	Language code (en, es, fr, etc.)
punctuate	boolean	Add punctuation (default: false)
diarize	boolean	Speaker identification (default: false)
smart_format	boolean	Format numbers, dates (default: false)

---

Response Format

Standard Response

Code
 
{
  "text": "Complete transcription text goes here",
  "segments": [
    {
      "start": 0.0,
      "end": 2.5,
      "text": "First segment of speech"
    },
    {
      "start": 2.5,
      "end": 5.8,
      "text": "Second segment of speech"
    }
  ],
  "language": "en",
  "duration": 5.8
}

With Diarization (Deepgram)

Code
 
{
  "text": "Complete transcription text",
  "segments": [
    {
      "start": 0.0,
      "end": 2.5,
      "text": "Hello, how can I help you?",
      "speaker": 0
    },
    {
      "start": 2.5,
      "end": 5.0,
      "text": "I need help with my account",
      "speaker": 1
    }
  ]
}

---

Error Handling

HTTP Status Codes

Code	Meaning	Solution
200	Success	Transcription returned
400	Bad Request	Check file format and parameters
401	Unauthorized	Verify API key
413	File Too Large	Reduce file size or split audio
415	Unsupported Format	Convert to supported format
429	Rate Limited	Implement backoff
500	Server Error	Retry with exponential backoff

Example with Retry Logic

Code
 
import time
from requests.exceptions import HTTPError

def transcribe_with_retry(
    file_path: str,
    max_retries: int = 3,
    api_key: str = "YOUR_API_KEY"
) -> str:
    """Transcribe with exponential backoff retry."""

    for attempt in range(max_retries):
        try:
            url = "https://api.slng.ai/v1/stt/slng/openai/whisper:large-v3"
            headers = {"Authorization": f"Bearer {api_key}"}

            with open(file_path, "rb") as f:
                files = {"file": f}
                response = requests.post(url, headers=headers, files=files, timeout=120)

            if response.status_code == 429:
                # Rate limited - wait and retry
                delay = (2 ** attempt) * 1000  # Exponential backoff
                print(f"Rate limited. Waiting {delay}ms...")
                time.sleep(delay / 1000)
                continue

            response.raise_for_status()
            return response.json()["text"]

        except HTTPError as e:
            if attempt == max_retries - 1:
                raise
            print(f"Attempt {attempt + 1} failed: {e}")
            time.sleep(2 ** attempt)

    raise Exception("Max retries exceeded")

# Usage
text = transcribe_with_retry("audio.mp3")

---

Use Cases

Meeting Transcription

Code
 
result = transcribe_file("team_meeting.mp3")

# Generate meeting summary
print("=== MEETING TRANSCRIPT ===\n")
print(result.text)
print("\n=== TIMELINE ===\n")
for segment in result.segments:
    minutes = int(segment.start // 60)
    seconds = int(segment.start % 60)
    print(f"[{minutes:02d}:{seconds:02d}] {segment.text}")

Podcast Episode

Code
 
# Transcribe long-form content
result = transcribe_file("podcast_episode.mp3", model="whisper", variant="large-v3")

# Create searchable transcript with timestamps
with open("episode_transcript.txt", "w") as f:
    f.write(f"Full Text:\n{result.text}\n\n")
    f.write("Timestamped Segments:\n")
    for segment in result.segments:
        f.write(f"[{segment.start:.1f}s] {segment.text}\n")

Customer Call Analysis

Code
 
# Use Deepgram for speaker separation
result = transcribe_file(
    "customer_call.wav",
    provider="deepgram",
    model="nova",
    variant="2"
)

# Separate by speaker
agent_text = []
customer_text = []

for segment in result.segments:
    if segment.get("speaker") == 0:
        agent_text.append(segment.text)
    else:
        customer_text.append(segment.text)

print("Agent:", " ".join(agent_text))
print("Customer:", " ".join(customer_text))