Azure Speech free tier: 5 audio hours is $3.00 on Sume STT

Microsoft's Azure Speech pricing page lists a free F0 tier with 5 audio hours free per month for standard speech-to-text and 0.5 million characters free per month for neural text-to-speech. On Sume STT at $0.01 per audio minute, those 5 hours would cost $3.00. The free tier is a fair way to prototype; the question is what you pay after you pass it.

What the page lists

The pricing page shows paid rates only after you pick a region and currency, so no paid Azure rate is quoted here.

The step after the free tier

Sume STT takes a public HTTPS audio_url and a duration_seconds hint of 1 to 600. For a five-hour backlog, split into 600-second pieces and submit them as separate jobs.

import os, requests

r = requests.post(
    "https://api.sume.com/v1/stt-1.0/transcribe",
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}",
             "Idempotency-Key": "stt-demo-001"},
    json={"audio_url": "https://example.com/call.mp3",
          "language_code": "en",
          "duration_seconds": 300},
)
print(r.status_code, r.json())

A sane prototype plan

Run your prototype inside the free allowance, keep the transcript format your app needs, then price a month of real audio on both services before you move volume.

Worked example

Use the free tier for what it is good at.

• Prototype: 5 hours of audio a month is 300 minutes, enough for dozens of test calls.
• Move to paid: on Sume, 300 minutes costs $3.00.
• TTS: 0.5 million characters is roughly 25 maximum-length Sume requests of 20,000 characters.

Checklist before you commit

Azure's paid rates need a region and currency. Pull your own figure before you plan a migration.

• Check Azure's region availability.
• Check output formats you need.
• Keep prototypes reproducible.