Google Cloud Text-to-Speech

Google Cloud Text-to-Speech is an API for turning text or SSML into speech audio in applications, devices, and voicebots. It is best suited to developers who need broad language coverage, controllable delivery, and production integration rather than a consumer reader. The main trade-off is a complex price menu in which model choice and text accounting materially affect cost.

Overview

The service offers more than 380 voices across more than 75 languages and variants, with REST and gRPC requests and streaming synthesis for low-latency conversations. That combination fits global products, contact-center voicebots, and accessible electronic program guides. Long-audio synthesis accepts up to 1 million bytes of input, useful for substantial passages, while streaming is better for interactive dialogue.

Billing must be enabled. Charges count input characters including spaces, newlines, and SSML tags, except the mark tag, so verbose markup can increase a bill. New Google Cloud customers can receive $300 in credits to try Text-to-Speech and other Google Cloud products. Google Cloud also maps security, privacy, and compliance controls to standards and subjects them to independent verification through certifications, attestations, and audit reports.

Key features

Voice selection and direction

Gemini-TTS handles single- or multispeaker speech and accepts natural-language prompts for style, accent, pace, tone, and emotional expression. This is a stronger fit for scripted characters, expressive assistants, and varied dialogue than a fixed voice catalogue, though Gemini models use a separate token-priced scheme. Chirp 3 instant custom voice can create a personalized voice model from as little as 10 seconds of audio and is available in more than 30 locales, making brand or character identity possible without a large recording session.

Controls, streaming, and output

The API allows pitch changes of up to 20 semitones, speaking rates up to four times faster or slower than normal, and volume gain from -96 dB to +16 dB. These ranges give applications room to tune narration for accessibility or production style. Output can be exported as MP3, LINEAR16, OGG_OPUS, MULAW, or ALAW, covering common files and telephony-oriented formats. Documentation includes guides, API references, quotas and limits, release notes, and support resources, which is valuable when an integration reaches service boundaries.

Pricing

Standard voices and WaveNet voices cost $0.000004 per character (US$4 per 1 million characters), with 0 to 4 million characters free; both show $0.00 per month. They are the clearest starting point for high-volume, conventional synthesis. Neural2 and Polyglot voices show $16.00 per month, include 1 million characters of free usage, and charge after that limit; they suit users who specifically need those voice families but require closer cost forecasting.

Chirp 3: HD voices show $0.00 per month and cost US$0.00003 per character (US$30 per 1 million), with 0 to 1 million characters free. Studio voices also show $0.00 per month, but cost US$0.00016 per character (US$160 per 1 million), with 0 to 1 million free. Instant custom voice shows $0.00 per month at US$0.00006 per character (US$60 per 1 million) and has no free usage limit. These tiers make expressive, studio, or personalized output substantially more expensive than the basic voices.

Gemini 2.5 Flash TTS and Gemini 2.5 Flash-Lite Preview TTS charge $0.50 per 1 million text tokens and $10.00 per 1 million audio tokens, with no free usage limit listed. Gemini 3.1 Flash TTS (Preview) charges $1.00 per 1 million text tokens and $20.00 per 1 million audio tokens, also with no free usage limit listed. Gemini 2.5 Pro TTS uses the same $1.00 input and $20.00 output token rates. Gemini 3.8 Flash TTS (Preview), through Dec 31, 2026, charges $0.50 per 1 million text tokens and $9.00 per 1 million audio tokens. The token model makes comparisons with per-character plans dependent on the actual text and audio generated.

Platforms

Google Cloud Text-to-Speech is available through an API and web access, with REST and gRPC integration. It supports commercial use and can return MP3, LINEAR16, OGG_OPUS, MULAW, or ALAW audio.

Who it's for

Choose it for software teams building multilingual voicebots, connected devices, accessible guides, or other products that need streaming, extensive controls, custom voices, and multiple export formats. It is less suitable for an individual seeking a simple read-aloud application, or for a project whose budget cannot tolerate model-specific rates and character or token metering.

Pros and cons

Pros

• Broad coverage: More than 380 voices across more than 75 languages and variants supports international deployments.
• Fine-grained delivery control: Pitch, speed, volume, and Gemini natural-language direction can shape speech for accessibility and characterful dialogue.
• Real-time and long-form options: Streaming serves conversations, while 1-million-byte long-audio input handles large passages.
• Flexible integration: REST, gRPC, five export formats, commercial use, and a $300 new-customer credit lower adoption friction for developers.

Cons

• Pricing is fragmented: Per-character, monthly, and token-priced models make budgeting dependent on the selected voice and input accounting.
• Billing setup is required: The service cannot be used without enabling billing, and spaces, newlines, and most SSML tags count toward charges.
• Premium voices cost more: Studio, HD, and custom options are much pricier than standard voices, while Instant custom voice has no free usage limit.

Alternatives

Fish Audio is a better fit when a freemium service with API, desktop, web, and self-hosted platforms matters; its free tier includes 8,000 credits monthly, up to seven minutes of generation, and 500 characters per generation. Choose Kokoro TTS instead for a free, self-hostable option with 82-million-parameter Apache-licensed weights. Speechify suits readers wanting a consumer-facing app, with a free plan offering 10 voices and 1.5x speed and a $29-per-month Premium plan with 60+ languages and 1,000+ voices. Typecast is worth considering for project-based media creation: its free plan allows three projects, 3,000 lifetime credits, 1 GB storage, and 720p exports. Cartesia Sonic offers API and web access with a free 20,000-credit monthly tier and a $5-per-month Pro tier. Chatterbox TTS is a free, self-hosted alternative under the MIT license. LOVO AI and Murf AI are freemium API/web services; Murf's free plan limits users to 10 projects, 10 minutes of generation, one editor, and no downloads.

For broader browsing, see AI Voice Generators, Game Voice Generators, and Text-to-Speech Software.

Verdict

Google Cloud Text-to-Speech is a strong choice for developers who need multilingual reach, live synthesis, adjustable or custom voices, and dependable API integration. Choose it for control and scale; look elsewhere if you want a simple consumer reader or a single, easily predictable subscription price.