dtelecom-stt

Python SDK for dTelecom real-time speech-to-text with x402 micropayments.

Pay-per-minute STT powered by Whisper and Parakeet, with automatic blockchain payments via USDC on Base or Solana.

Install

pip install dtelecom-stt

Quick Start

import asyncio
from dtelecom_stt import STTClient

async def main():
    # EVM wallet (Base)
    client = STTClient(private_key="0x...")
    # Or Solana wallet — detected automatically
    # client = STTClient(private_key="base58...")

    async with client.session(minutes=5, language="en") as stream:
        async for t in stream.transcribe_file("meeting.wav"):
            print(f"[{t.start:.1f}s] {t.text}")

asyncio.run(main())

Real-Time Streaming

import asyncio
from dtelecom_stt import STTClient

async def main():
    client = STTClient(private_key="0x...")  # or Solana base58 key

    async with client.session(minutes=5, language="en") as stream:
        # Callback-based
        stream.on_transcription(lambda t: print(t.text))

        # Send audio chunks (PCM16, 16kHz, mono)
        await stream.send_audio(pcm_bytes)

        # Or iterate asynchronously
        async for t in stream.transcriptions():
            print(t.text)

asyncio.run(main())

Auto-Extend Sessions

Sessions automatically buy more time when running low (enabled by default):

# 30-minute session that auto-extends
async with client.session(minutes=30, language="en") as stream:
    # When <60s remaining, SDK buys 5 more minutes automatically
    async for t in stream.transcriptions():
        print(t.text)

# Disable auto-extend
async with client.session(minutes=5, auto_extend=False) as stream:
    ...

Audio Format

The server expects PCM16, 16kHz, mono audio. Convert with ffmpeg:

ffmpeg -i input.mp3 -ar 16000 -ac 1 -acodec pcm_s16le output.wav

Pricing

info = await client.pricing()
print(f"${info.price_per_minute_usd}/min ({info.currency} on {info.network})")

Current pricing: $0.005/min (USDC on Base or Solana).

API Reference

STTClient(private_key, url=None)

Main client. Default URL: https://x402stt.dtelecom.org.

• EVM key (hex, 0x-prefixed): pays with USDC on Base

• Solana key (base58): pays with USDC on Solana

• session(minutes=5, language="en", auto_extend=True) — Create a paid session (async context manager)

• pricing() — Get pricing info

• health() — Check server health

Stream

Returned by client.session(). Async context manager.

• send_audio(data: bytes) — Send raw PCM16 audio
• transcriptions() — Async iterator of Transcription objects
• transcribe_file(path) — Stream a WAV file and yield transcriptions
• on_transcription(callback) — Register callback for transcriptions
• close() — Close the stream

Transcription

• text: str — Transcribed text
• start: float | None — Start time in seconds
• end: float | None — End time in seconds
• confidence: float | None — Confidence score
• is_final: bool — Whether this is a final transcription

Supported Languages

25 languages via Parakeet-TDT (fast) with Whisper fallback:

English, Russian, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Czech, Romanian, Hungarian, Greek, Turkish, Ukrainian, Swedish, Norwegian, Danish, Finnish, Catalan, Croatian, Lithuanian, Slovenian, Latvian, Estonian.

Error Handling

from dtelecom_stt import PaymentError, SessionExpiredError, ConnectionError

try:
    async with client.session(minutes=5) as stream:
        async for t in stream.transcriptions():
            print(t.text)
except PaymentError:
    print("Payment failed — check wallet balance")
except SessionExpiredError:
    print("Session time ran out")
except ConnectionError:
    print("Cannot connect to server")

License

MIT