Speech-to-Text API Pricing & Credit Rates

Transparent, pay-per-second usage billing for speech recognition, automated transcription, and real-time streaming audio.

Credit Consumption Rates

YourVoic Speech-to-Text uses a transparent credit model. You only pay for the exact audio duration processed, measured down to the second.

Model FamilyEngines / ModelsCredits/SecondCredits/MinuteCredits/Hour
Cipher Fastcipher-fast2 credits120 credits7,200 credits
Cipher Maxcipher-max2 credits120 credits7,200 credits
Lucid Monolucid-mono3 credits180 credits10,800 credits
Lucid Multilucid-multi3 credits180 credits10,800 credits
Lucid Agentlucid-agent3 credits180 credits10,800 credits
Lucid Litelucid-lite3 credits180 credits10,800 credits

Batch Processing Billing Mechanics

📊 Deterministic Calculation: For file uploads, audio duration is read directly from file headers prior to dispatching neural inference. Credits are reserved up front, and failed jobs are never charged.
  • Exact duration accounting: A 47-second audio file on cipher-fast costs exactly 94 credits (47s × 2 credits/s). We never round up to arbitrary 15-second or 1-minute blocks.
  • Included features: Speaker diarization, punctuation, casing, smart number formatting, and word timestamps are included at zero extra charge.
  • Failed job guarantee: If an audio file cannot be decoded or encounters a worker fault, credits are automatically refunded to your balance instantly.

Real-time Streaming Session Billing

🎙️ Live Stream Usage: Real-time WebSocket sessions are billed strictly on audio stream duration transmitted at 3 credits per second.
  • Startup balance: A minimum balance of 60 credits is required to initiate a WebSocket session.
  • Continuous verification: Active balance is validated every 30 seconds during an ongoing call.
  • Graceful disconnect: If credits deplete during a session, the server flushes final transcripts before closing the socket.

Programmatic Credit & Balance Checking

Developers can monitor remaining credits programmatically before queuing high-volume batches using the Account Info API or by inspecting response headers.

1. Check Account Balance (Python)

import requests

url = "https://yourvoic.com/api/v1/user/credits"
headers = {"X-API-Key": "YOUR_API_KEY"}

response = requests.get(url, headers=headers)
data = response.json()

if response.status_code == 200:
    balance = data["credits"]["remaining"]
    print(f"Available credits: {balance:,}")
    # Estimate minutes available for cipher-fast (120 credits/min)
    available_minutes = balance // 120
    print(f"Estimated transcription time: {available_minutes} minutes")
else:
    print(f"Error checking balance: {data.get('message')}")

2. Check Balance via cURL

curl -X GET "https://yourvoic.com/api/v1/user/credits" \
  -H "X-API-Key: YOUR_API_KEY"

3. Account Credits Response JSON (200 OK)

{
  "success": true,
  "account_id": "usr_78a1bc40d29",
  "plan": "developer_pro",
  "status": "active",
  "credits": {
    "total_monthly_allowance": 500000,
    "consumed_this_cycle": 142800,
    "remaining": 357200,
    "bonus_credits": 5000
  },
  "billing_cycle": {
    "started_at": "2026-10-01T00:00:00Z",
    "renews_at": "2026-11-01T00:00:00Z"
  },
  "concurrency_limit": 10
}

Response Headers for Real-time Accounting

Every batch transcription response returns audit headers allowing your application to log usage without separate balance queries:

HTTP/1.1 200 OK
Content-Type: application/json
X-Audio-Duration: 42.8
X-Credits-Used: 86
X-Credits-Remaining: 357114
X-RateLimit-Limit: 120
X-RateLimit-Remaining: 119

Real-World Cost Estimation Scenarios

Application / Use CaseMonthly Audio VolumeRecommended ModelMonthly Credits
Podcasts & Creator Content20 episodes × 45 min (15 hrs)cipher-fast108,000 credits
Call Center QA / Auditing1,000 calls × 6 min (100 hrs)cipher-max720,000 credits
Live AI Voice Agent50 hours live audiolucid-agent540,000 credits
Lecture & Webinar Captions10 sessions × 90 min (15 hrs)lucid-multi162,000 credits

Frequently Asked Questions (FAQ)

How are fractions of a second billed?

Audio duration is calculated to the millisecond. For example, 10.4 seconds of audio processed by cipher-fast (2 credits/sec) requires 21 credits (rounded to the nearest integer credit), avoiding penalty fees.

Do credits expire at the end of the month?

Monthly plan credits refresh at your recurring billing date. Top-up credit packs purchased separately never expire and roll over indefinitely until consumed.

Are silent pauses in audio charged?

Billing is based on total submitted audio file duration. To minimize credit consumption on long multi-hour recordings with extended silence, we recommend running client-side voice activity detection (VAD) prior to streaming.

How can I purchase additional credits?

Visit your Account Dashboard to purchase credit packs or upgrade to a higher tier plan with volume discounts and custom concurrency limits.

Enterprise Solutions

For organizations processing over 1,000 hours of speech per month, custom enterprise packages offer dedicated GPU workers, on-premise deployment options, HIPAA/GDPR business associate agreements, and custom acoustic model tuning.