Short answer: ElevenLabs is the clearest paid creator benchmark because it publishes credit-based plans, voice cloning access, commercial licensing, and API access in one place. Amazon Polly and Google Cloud Text-to-Speech are usually easier to model for high-volume API costs. Fliki, Murf, LOVO, Descript, Speechify, and WellSaid Labs make more sense when the workflow needs a browser studio, video editing, brand narration, or creator-friendly production tools rather than raw character pricing.
This page is a pricing and quota reference for AI voice buyers, creators, and developers comparing text-to-speech tools. Prices were checked from official pricing pages or official product pages in July 2026. When a vendor's official pricing app is JavaScript-gated, the table uses a clearly labeled third-party pricing source only where the plan name, price, and quota are specific enough to verify.
For product-specific recommendations, start with the broader AI voice and TTS comparison hub, then use this page to sanity-check cost. If you are choosing for a specific workflow, also compare the best voice cloning software, best TTS for audiobooks, best TTS for podcasts, best text-to-speech API for developers, best free ElevenLabs alternatives, and ElevenLabs review.
Quick Recommendation
If you want one premium AI voice platform with realistic output, voice cloning, and an API, start with ElevenLabs. If you need predictable infrastructure pricing, compare Amazon Polly and Google Cloud Text-to-Speech. If the job is voiceover inside a production workflow, compare Murf AI, Fliki, Descript, Speechify, and WellSaid Labs.
For most small teams, the buying decision should not be "which tool is cheapest per character?" It should be "which tool gives me the right output with the fewest retakes, editing steps, licensing questions, and export limits?"
TTS Pricing And Quota Table
| Tool | Starting Price | Free Tier | Paid Quota / Cost Model | Voices / Languages | Commercial Use | API | Voice Cloning | Best Fit |
|---|---|---|---|---|---|---|---|---|
| ElevenLabs | $6/mo Starter | 10k credits/mo | 30k Starter, 121k Creator, 600k Pro, 1.8M Scale, 6M Business credits | Varies, broad voice library | Starter+ includes commercial license | Yes | Yes | Realistic voices, cloning, creator and API workflows |
| Murf AI | $29/mo Creator | Free plan | Creator includes 2 hours/mo voice generation; Business includes 8 hours/mo | 120+ voices, 20+ languages | Paid plans | Yes | Enterprise/add-on | Browser voiceover studio and business narration |
| LOVO AI | See official pricing | 14-day Pro trial | Voice generation hours vary by plan | 500+ voices, 100+ languages | User owns generated content rights per LOVO FAQ | Yes | Yes | Expressive creator voiceover and dubbing |
| Fliki | Free; paid plans with annual discount | 5 free audio/video minutes/mo | Standard includes 180 minutes/mo or 2160 annual credits | 1300+ voices, 80+ languages | Subscription plans include commercial license | Verify current access | Yes | Script-to-video and multilingual social content |
| Speechify | $29/mo Premium | Free reader tier | Consumer Premium; Studio/API priced separately | 1000+ voices, 60+ languages | Check Studio/API terms | Yes | Yes | Reading, accessibility, Studio, and API paths |
| Play.ht / PlayAI | Official homepage says service shut down | Not recommended | Not applicable | Historic copy lists 800+ voices, 30+ languages | Not recommended for new purchases | Legacy docs exist | Historic feature | Legacy migration checks |
| WellSaid Labs | $10/mo Starter, billed annually | Trial with 3 downloaded minutes/mo | 240 downloaded min/year Starter, 2160 Pro, 2880 Business/user | Varies by plan | Paid studio plans | Confirm with sales | Custom workflows | Brand-safe business narration |
| Descript | $24/person/mo Creator annual, $35 monthly | Free plan | Hobbyist has 10 media hours/mo and 400 AI credits/mo | Stock voices plus custom voice clones | Creator/business workflows | Yes | Yes | Podcast/video editing and AI speech repair |
| Amazon Polly | $4 per 1M Standard chars | 5M Standard chars/mo, 1M Neural chars/mo for first 12 months | $16 Neural, $30 Generative, $100 Long-Form per 1M chars | Varies by engine/region | AWS terms | Yes | No self-serve cloning | High-volume backend TTS |
| Google Cloud TTS | $4 per 1M Standard/WaveNet chars after free tier | 4M Standard/WaveNet chars, 1M Neural2/Chirp depending on model | $16 Neural2, $30 Chirp 3 HD, $60 Instant custom voice per 1M chars | Varies by model | Google Cloud terms | Yes | Instant custom voice listed separately | Developer TTS and cloud applications |
| OpenAI TTS | Token-based audio generation pricing | No standing free tier listed | Realtime audio generation priced per 1M tokens; some audio models priced per minute | 13 built-in Speech API voices | OpenAI API terms | Yes | No self-serve cloning | Developer speech generation and realtime apps |
| Azure Speech | Pay-as-you-go, region dependent | 0.5M Neural TTS chars/mo on Free F0; web/container Standard lists 5M chars/mo | Paid pricing varies by region and voice type | Varies by region/model | Azure terms | Yes | Custom neural voice approval path | Enterprise Azure deployments |
How To Read This Table
The cheapest row is not always the cheapest workflow. A tool with a lower per-character price can cost more if it requires extra editing, forces repeated generations, limits commercial use, or lacks the exact voice style you need. For example, an audiobook creator should care about long-form consistency and chapter-level revision time, while a SaaS developer should care about API latency, concurrency, predictable metering, and error handling.
Subscription tools and cloud APIs also measure usage differently. ElevenLabs uses monthly credits shared across products. WellSaid Labs talks about downloaded finished audio minutes. Fliki uses credits tied to generated or exported audio/video. Amazon Polly and Google Cloud Text-to-Speech are closer to infrastructure pricing, with per-character rates by voice class. OpenAI's newer voice stack is token-based for realtime audio models, which makes it powerful but less intuitive for buyers who think in characters or minutes.
The safest way to compare real cost is to run the same script through two or three shortlisted tools. Use a representative script, not a perfect demo sentence. Include names, numbers, punctuation, pauses, product terms, and a sentence that requires emotion. Count not only the listed price, but also how many times you had to regenerate, edit, export, or clean up the audio.
Notes By Tool
ElevenLabs
ElevenLabs is the clearest premium benchmark because its pricing page lists the Free, Starter, Creator, Pro, Scale, Business, and Enterprise tiers with monthly credit pools. The important buyer detail is that commercial licensing starts on Starter, while Professional Voice Cloning starts higher. That makes it stronger for creators who need realistic narration, custom voices, and API access in one ecosystem. For deeper workflow context, see ElevenLabs alternatives and ElevenLabs vs Murf.
Murf AI
Murf is less about raw API character pricing and more about a controlled voiceover studio. Its official pricing app is JavaScript-gated in static checks, but G2's Murf pricing page lists Creator at $29/month with 2 hours of voice generation per month and Business at $99/month with 8 hours per month. Treat those figures as a July 2026 planning baseline and confirm the live Murf pricing page before buying. Murf remains useful when the buyer needs voiceover editing, collaboration, brand-safe training content, marketing narration, or a workflow that non-developers can operate. The API path exists, but most content buyers will compare Murf on studio usability and export workflow first.
LOVO AI
LOVO's public site emphasizes Genny, 500+ voices, 100+ languages, voice cloning from a short sample, API availability, a 14-day Pro trial, and user ownership of generated content. Exact plan pricing was not statically exposed during this check, so the table keeps LOVO's price conservative instead of importing an unverified number. LOVO can still be worth covering because it has search demand and fits expressive creator voiceover, dubbing, and multilingual video workflows. Since its affiliate program is unavailable right now, commercial CTAs should usually point buyers toward ElevenLabs, Murf, or Fliki where appropriate.
Fliki
Fliki is strongest when the user wants text-to-video, social content, multilingual voiceover, and creator-friendly exports. Its official pricing page states that users can generate 5 minutes of free audio/video per month, that the Standard plan includes 180 minutes of credits, and that subscription plans include a commercial license. Fliki also lists 1300+ voices and 80+ languages. That makes it a different buying decision from Amazon Polly or Google Cloud TTS: the value is production workflow, not just raw TTS cost.
Speechify
Speechify has multiple product paths, so buyers should separate the consumer reader plan from Studio and API use. The public pricing page lists a free reader tier and Premium at $29/month, with 1000+ high-quality voices and 60+ languages. For a content business, the key question is whether you are buying a reader, a creator voiceover tool, or developer access. Do not assume the consumer Premium plan covers every commercial Studio/API workflow without checking the relevant product terms.
Play.ht / PlayAI
Play.ht is included because many older comparison pages and searchers still ask about it. Its official homepage includes a shutdown notice, so it should not be treated as a normal active buying option. Historic homepage copy still references a large voice library, API, and voice cloning, but the practical recommendation is migration research. Current buyers should compare PlayHT alternatives instead of planning a new Play.ht deployment.
WellSaid Labs
WellSaid Labs is positioned for polished business narration rather than bargain raw TTS. Its official pricing page lists Starter at $10/month billed annually, Pro at $33/month billed annually, Business at $160/month/user billed annually, and Enterprise as custom. The key quota is downloaded finished audio minutes: 240 minutes/year on Starter, 2160 on Pro, and 2880 per user on Business. That makes WellSaid a better fit for brand narration, training, and controlled business voice work than casual high-volume API experimentation.
Descript
Descript is not a pure TTS tool. It is a video, podcast, transcription, and AI editing platform that includes AI Speech, stock voices, and custom voice clones. Its official pricing page lists Creator at $24/person/month annually or $35 monthly, and references media hours, AI credits, AI Speech, and API/MCP access. It is best compared when the buyer also needs editing, transcription, filler-word cleanup, screen recording, or voice repair. For a voice-only decision, compare it against purpose-built TTS tools separately.
Amazon Polly
Amazon Polly is the cleanest infrastructure-style comparison row. AWS lists $4 per 1 million characters for Standard voices, $16 per 1 million for Neural voices, $30 per 1 million for Generative voices, and $100 per 1 million for Long-Form voices. The free tier includes millions of characters for new users and varies by voice type. Polly is not the obvious choice for creator-style voice cloning, but it is strong when teams need predictable backend TTS, AWS integration, and usage-based billing.
Google Cloud Text-to-Speech
Google Cloud Text-to-Speech is also infrastructure-first. Its pricing page lists legacy Standard/WaveNet rates, Neural2 rates, Chirp 3 HD, Gemini TTS, and Instant custom voice pricing separately. The table uses the clearest buyer-facing rates: $4 per 1M Standard/WaveNet characters after the free tier, $16 per 1M Neural2 characters, $30 per 1M Chirp 3 HD characters, and $60 per 1M Instant custom voice characters. It is a strong fit for developers comparing cloud APIs, especially if they already use Google Cloud.
OpenAI TTS
OpenAI's speech generation stack is developer-first. The speech docs list `gpt-4o-mini-tts` as the newest and most reliable text-to-speech model and list 13 built-in voices. The pricing page groups realtime and audio generation models under token pricing rather than a simple per-character plan. That means buyers should model cost from actual app usage, especially for realtime voice agents. OpenAI is best when prompt-controlled voice, realtime interaction, and broader AI app integration matter more than a traditional voiceover studio.
Azure Speech
Azure Speech is best for organizations already committed to Microsoft cloud, governance, and enterprise deployment. Its pricing page lists a Free F0 tier with 0.5 million Neural Text to Speech characters per month, and a separate web/container table listing 5 million Standard characters free per month. Paid pricing can vary by region and voice type, so this page avoids a universal number. Azure is worth evaluating for enterprise custom voice projects, but custom neural voice access generally requires approval and policy compliance.
Cost Model Checklist
Before choosing a TTS platform, run this checklist:
- Is the plan metered by characters, credits, generated minutes, downloaded minutes, or tokens?
- Does the free tier allow commercial use, or only testing?
- Does voice cloning require a paid plan, business approval, or consent workflow?
- Does API access ship on self-serve plans, or does it require sales?
- Are retries and regenerations charged?
- Can you export clean audio at the quality your workflow needs?
- Does the tool keep pronunciation, pacing, and emotional tone stable across long scripts?
- Can the vendor support your compliance needs if the voice is used in ads, education, healthcare, finance, or customer support?
Recommendations By Buyer Type
Creators making YouTube videos, podcasts, short courses, and social clips should start with ElevenLabs, Murf, Fliki, Descript, or Speechify Studio. The best choice depends on whether the main bottleneck is voice realism, editing, video assembly, or reading/listening workflows.
Developers should start with ElevenLabs, Amazon Polly, Google Cloud Text-to-Speech, OpenAI TTS, Azure Speech, or Speechify API. The right choice depends on latency, language coverage, compliance, region, voice quality, and whether the app needs voice cloning.
Teams producing brand narration should compare WellSaid Labs, ElevenLabs, Murf, and Descript. These tools are easier to evaluate by final output quality and approval workflow than by raw per-character pricing.
Budget-conscious users should read best free ElevenLabs alternatives before choosing. Free plans are useful for testing voices, but they often restrict commercial use, export quality, volume, or voice cloning.
FAQ
What is the cheapest text-to-speech API?
Amazon Polly and Google Cloud Text-to-Speech are usually the easiest low-cost API options to model because they publish per-character pricing. Amazon Polly lists Standard voices at $4 per 1 million characters, while Google Cloud lists Standard/WaveNet legacy voices at $4 per 1 million characters after the free tier. The cheapest practical API can still depend on voice quality, language, latency, and retries.
Which TTS tool has the best free tier?
The best free tier depends on what you need to test. ElevenLabs is useful for testing premium AI voice quality with a credit pool. Fliki gives a small monthly audio/video allowance. Speechify has a free reader tier. Amazon Polly, Google Cloud, and Azure offer cloud free tiers for limited API usage. Free does not always mean commercial-ready, so check rights before publishing client work.
Does every TTS pricing plan include commercial use?
No. Some tools reserve commercial rights for paid plans, some bundle it with subscription tiers, and cloud APIs rely on platform terms. ElevenLabs lists commercial licensing from Starter and above. Fliki says subscription plans include a commercial license. LOVO states users own generated content rights. For any client, ad, audiobook, or monetized YouTube use, confirm the current terms before purchase.
Which TTS tools support voice cloning?
ElevenLabs, LOVO, Fliki, Speechify creator products, Descript, and some WellSaid/custom enterprise workflows support voice cloning or custom voice features. Google Cloud lists Instant custom voice separately, while Azure offers custom neural voice through an approval path. Amazon Polly is not a self-serve voice cloning product in the same creator-tool sense.
Should developers choose a cloud API or a creator TTS tool?
Developers should choose a cloud API when predictable metering, infrastructure integration, and scale matter most. Amazon Polly, Google Cloud Text-to-Speech, Azure Speech, and OpenAI TTS fit that pattern. Creator TTS tools make more sense when a product team needs higher-touch voice quality, cloning, studio controls, or non-technical staff reviewing audio before release.
How often should TTS pricing be updated?
For a public comparison page, update pricing at least quarterly and whenever a major vendor changes plan names, free tier limits, API pricing, or commercial-use terms. This page uses `content-pipeline/data/tts-pricing.json` as the structured source so the table can be refreshed without rewriting the whole article.
Short answer: use ElevenLabs as the premium creator benchmark, Amazon Polly and Google Cloud as infrastructure price anchors, and workflow tools like Murf, Fliki, Descript, Speechify, LOVO, and WellSaid when editing, video, brand safety, or voice production matters more than raw character cost.