Google now offers speech AI through multiple different services, each with its own pricing model. Whether you’re building a voice assistant, transcribing podcasts, or generating audiobook narrations, this guide maps every option and its exact cost.
🗣️ The Google Speech AI Landscape
Google’s speech capabilities are split across three distinct services:
| Service | What It Does | Pricing Model |
|---|---|---|
| Gemini TTS | Text → Natural speech | Per token |
| Gemini Live API | Real-time bidirectional voice | Per token (audio) |
| Google Cloud STT | Audio → Text transcription | Per minute |
| Gemini Audio Understanding | Audio → Analysis/summary | Per token |
Let’s break each one down.
🎙️ Gemini TTS Models — Text-to-Speech
Gemini 3.1 Flash TTS (Preview)
The latest and most capable TTS model. Supports 70+ languages, multi-speaker synthesis, and granular control via natural language audio tags.
Model ID: gemini-3.1-flash-tts-preview
| Cost Type | Price per 1M Tokens |
|---|---|
| Input (text) | $1.00 |
| Output (audio) | $20.00 |
| Batch Input | $0.50 |
| Batch Output | $10.00 |
💡 Token-to-seconds conversion: Audio output tokens correspond to 25 tokens per second. This means 1 million output tokens = ~11 hours of audio.
Real-World Cost Example
Generate a 10-minute podcast narration (~6,000 words input, 600 seconds of audio):
| Cost Component | Calculation | Amount |
|---|---|---|
| Input tokens (~8,000) | 8K × $1.00/1M | $0.008 |
| Output tokens (600s × 25) | 15,000 × $20.00/1M | $0.30 |
| Total | ~$0.31 |
At 31 cents for 10 minutes of high-quality speech, Gemini TTS is extremely competitive.
Gemini 2.5 Flash TTS (Preview)
A more cost-efficient option, optimized for lower latency.
Model ID: gemini-2.5-flash-preview-tts
| Cost Type | Price per 1M Tokens |
|---|---|
| Input (text) | $0.15 |
| Output (audio) | $6.00 |
| Batch Input | $0.075 |
| Batch Output | $3.00 |
🎯 Best for: High-volume TTS at lower cost. At $6/M output tokens, a 10-minute narration costs only ~$0.09 — 3× cheaper than 3.1 Flash TTS.
Gemini 2.5 Pro TTS (Preview)
The premium TTS option — more natural outputs and easier-to-steer prompts.
Model ID: gemini-2.5-pro-preview-tts
| Cost Type | Price per 1M Tokens |
|---|---|
| Input (text) | $1.25 |
| Output (audio) | $20.00 |
| Batch Input | $0.625 |
| Batch Output | $10.00 |
📊 TTS Model Comparison at a Glance
| Model | Input/1M | Output/1M | 10-min Cost | Best For |
|---|---|---|---|---|
| 2.5 Flash TTS | $0.15 | $6.00 | ~$0.09 | Budget narration, IVR |
| 3.1 Flash TTS | $1.00 | $20.00 | ~$0.31 | Natural conversation, apps |
| 2.5 Pro TTS | $1.25 | $20.00 | ~$0.31 | Premium voice quality |
🔴 Gemini Live API — Real-Time Voice
For real-time, bidirectional audio conversations (think: voice assistants, customer service bots).
Model ID: gemini-3.1-flash-live-preview
| Cost Type | Price per 1M Tokens |
|---|---|
| Text Input | $1.00 |
| Audio Input | $1.00 |
| Text Output | $6.00 |
| Audio Output | $20.00 |
Key Details
- Audio tokens: 25 tokens per second of audio (both input and output)
- Real-time streaming: Low-latency, designed for conversational agents
- Multimodal: Supports audio + video input for “see and speak” applications
- Acoustic awareness: Detects tone, emotion, and ambient sounds
⚠️ Cost warning: A 1-hour voice conversation would cost approximately $1.80 in audio input + $72 in audio output. Design your agent to keep responses concise!
🎧 Gemini Audio Understanding — Speech-to-Analysis
Instead of traditional STT, Gemini can directly analyze, summarize, and reason about audio content using its multimodal capabilities.
| Model | Input/1M Tokens | Output/1M Tokens |
|---|---|---|
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 |
| Gemini 3 Flash | $0.50 | $3.00 |
| Gemini 3.1 Pro | $2.00 | $12.00 |
Use this when you need more than transcription — summarization, sentiment analysis, meeting action items, translation, or Q&A about audio content.
📝 Google Cloud Speech-to-Text (Chirp) — Traditional STT
For dedicated, high-accuracy transcription, Google Cloud’s STT API remains the best option.
| Model | Price per Minute | Free Tier |
|---|---|---|
| Chirp (Standard) | $0.016 | 60 min/month |
| Chirp (High Volume 500K+ min) | $0.012 | — |
| Chirp (High Volume 2M+ min) | $0.004 | — |
Important Billing Details
- Billed in 15-second increments (rounds up)
- Supports 125+ languages with auto-detection
- $300 free credits for new Google Cloud accounts
🧮 Full Cost Comparison: All Speech Services
Scenario: Process 1 hour of audio content
| Task | Best Service | Estimated Cost |
|---|---|---|
| Transcribe audio → text | Google Cloud STT (Chirp) | $0.96 |
| Summarize meeting audio | Gemini 3 Flash (audio input) | ~$1.50 |
| Generate 1-hour audiobook | Gemini 2.5 Flash TTS | ~$5.40 |
| Generate 1-hour audiobook | Gemini 3.1 Flash TTS | ~$18.00 |
| Real-time voice agent (1 hour) | Gemini Live API | ~$73.80 |
✅ Key Takeaways
- Gemini 2.5 Flash TTS is the cheapest option at $0.09 per 10 minutes of speech
- Gemini 3.1 Flash TTS offers the best quality-to-cost ratio for natural speech
- Live API is designed for real-time agents — but costs add up fast for long conversations
- Google Cloud STT is still the cheapest for pure transcription at $0.016/minute
- Gemini Audio Understanding is the smart choice when you need analysis, not just transcription
- Batch API saves 50% on all TTS models for non-urgent workloads
Get Started
Try Gemini TTS in Google AI Studio or check the official pricing page for the latest rates.
Prices are current as of May 2026. Always verify with Google’s official documentation before production deployment.
