What Happened to Play.ht — and What to Use Instead
Play.ht is gone. Meta acquired the company on July 12, 2025, absorbed the 35-person team into its Superintelligence Labs division, and shut the platform down on December 31, 2025. The API actually went dark on July 26 — weeks ahead of the announced timeline — breaking integrations overnight for thousands of developers and SaaS builders. All user data (accounts, voice clones, saved audio, API keys) was deleted with no export tools provided.
If you're a former Play.ht user looking for a replacement, here's the short version: for voice quality, ElevenLabs ($5/mo starter) is the closest upgrade. For budget-conscious teams, Fish Audio offers better blind-test results at $15/1M characters — 6x cheaper than Play.ht's old pricing. For developers who want zero vendor lock-in, Chatterbox is fully open-source under MIT license.
The Full Shutdown Timeline
| Date | Event |
|---|---|
| Jul 11, 2025 | Bloomberg reports Meta acquisition |
| Jul 12, 2025 | Meta confirms acqui-hire of entire 35-person team |
| Jul 26, 2025 | API stops accepting requests — 2 weeks early, no warning |
| Aug 2025 | Platform stops accepting new sign-ups |
| Dec 31, 2025 | Full shutdown — all data deleted per Meta privacy policy |
The worst part: Meta provided no migration tools. If you didn't manually download your generated audio and voice clone data before the cutoff, it's gone permanently. This affected everyone from solo podcasters to SaaS companies that had built voice features on Play.ht's API.
What Play.ht Users Actually Need in a Replacement
Play.ht's strengths were straightforward: decent voice cloning, a simple REST API, reasonable pricing ($39/mo for the Pro plan), and multi-language support. Any replacement needs to match at least those basics. But the TTS landscape has changed dramatically since Play.ht launched — most alternatives now exceed what Play.ht offered.
The key factors to evaluate:
- API compatibility — How much code do you need to rewrite? Most modern TTS APIs use similar REST patterns, but the response formats and streaming behavior differ.
- Voice cloning — If you had custom voices on Play.ht, you'll need to re-record reference audio and re-clone. The good news: newer services clone from as little as 3 seconds.
- Pricing model — Play.ht charged per-character on API and flat monthly for the web app. Alternatives range from per-character to per-minute to completely free.
- Vendor lock-in risk — After getting burned by Meta's shutdown, many former Play.ht users now prefer open-source options or services with data export guarantees.
1. ElevenLabs — Best Overall Replacement ($5–$99/mo)
ElevenLabs is the default recommendation for former Play.ht users, and for good reason. The voice quality is measurably better — ElevenLabs consistently ranks in the top 3 on the Artificial Analysis TTS Arena. The Starter plan at $5/month is cheaper than Play.ht's old $39/mo Pro plan and includes voice cloning from 30 seconds of audio.
The API is well-documented and supports WebSocket streaming for real-time applications. If you were using Play.ht for podcast production, audiobook narration, or voiceovers, ElevenLabs handles all of those with 1,000+ voices across 32 languages.
Downsides: the per-character pricing gets expensive at scale. At 1M characters/month, you're looking at $99/mo (Pro plan) vs Fish Audio's $15/1M. For detailed cost breakdowns at every usage level, see our ElevenLabs pricing guide.
ElevenLabs vs Play.ht (Former)
2. Fish Audio — Best Value for API Users ($15/1M Characters)
Fish Audio S2 Pro ranked #1 in blind A/B tests with a Bradley-Terry score of 3.07 — nearly 1.7x the next best model. And it costs $15 per million characters. To put that in perspective: Play.ht's API pricing worked out to roughly $90–$165 per million characters depending on your plan. Fish Audio is 6–11x cheaper with better quality.
The platform supports 80+ languages and offers voice cloning from short reference samples. The API follows standard REST patterns, so migration from Play.ht is mostly a matter of changing endpoints and adjusting for Fish Audio's UTF-8 byte-based billing (important gotcha for CJK languages — characters cost 3x more due to multi-byte encoding). Full pricing analysis in our Fish Audio pricing guide.
Fish Audio also offers self-hosting under Apache 2.0 license. If vendor lock-in is now a dealbreaker for you (understandable after Play.ht), you can run the model on your own infrastructure and never worry about another acqui-hire pulling the rug. See our Fish Audio vs ElevenLabs comparison for a detailed head-to-head.
3. Murf AI — Free 6-Month Offer for Play.ht Users ($23/mo After)
Murf AI launched a dedicated migration program for displaced Play.ht users: a free 6-month subscription matching or upgrading your existing Play.ht plan. If you're reading this and haven't claimed that offer yet, it may still be available at murf.ai/play-ai-migration.
Murf is the most similar to Play.ht in terms of interface and workflow. It's a web-based studio with 120+ voices across 20 languages, a video editor for syncing voiceovers, and API access via the Falcon engine. The Gen-3 voice engine is solid for business presentations, e-learning, and marketing content.
The honest downside: Murf's voice quality has fallen behind newer competitors. In blind comparisons, ElevenLabs and Fish Audio both sound noticeably more natural. And the pricing — $23/mo on annual billing, $39/mo monthly — is no longer competitive for what you get. The Business plan is $66/mo, not $33/mo as Murf's pricing page implies (see our Murf AI review for the corrected numbers). Still, that free 6-month offer is hard to beat for a bridge while you evaluate other options.
4. Amazon Polly — Best for AWS Developers ($4–$16/1M Characters)
If you were using Play.ht's API for programmatic TTS and your infrastructure already runs on AWS, Amazon Polly is the pragmatic choice. Standard voices cost $4 per million characters — the cheapest managed API on the market. Neural voices are $16/1M for significantly better quality. And the free tier is generous: 5M standard characters + 1M neural characters per month for 12 months.
The trade-off is voice variety and naturalness. Polly's voices are functional but not in the same league as ElevenLabs or Fish Audio for creative content. They're great for IVR systems, accessibility features, and notification audio. For detailed pricing across all four engine types, see our Amazon Polly pricing guide.
5. OpenAI TTS — Best for ChatGPT Ecosystem Users ($15/1M Characters)
If you're already using the OpenAI SDK for other AI features, adding TTS is a single API call away. At $15 per million characters with no monthly commitment, it's simpler than Play.ht's subscription model — you pay only for what you use. The voices are limited to a handful of presets (alloy, echo, fable, onyx, nova, shimmer, ash, ballad, coral, sage, verse), but they're high quality.
No voice cloning, no SSML control, no custom voice creation. OpenAI TTS is intentionally simple. If Play.ht's voice cloning was important to you, this isn't the right replacement. But if you just need clean, natural-sounding speech output in your app, it's one of the easiest migrations. Full breakdown in our OpenAI TTS pricing guide.
6. Chatterbox — Best Open-Source Option (Free, MIT License)
After watching Play.ht disappear into Meta's corporate machinery, a lot of former users swore off proprietary TTS entirely. If that's you, Chatterbox is the answer. Built by Resemble AI and released under the MIT license, Chatterbox won 65.3% of blind tests against ElevenLabs — and it's completely free.
The model handles voice cloning from just 5 seconds of reference audio and supports native paralinguistic tags ([laugh], [cough], [sigh]) for more expressive output. It runs locally on your hardware — no API calls, no monthly bills, no risk of another shutdown pulling your voices away. Read our full Chatterbox review for setup details.
Other strong open-source options: Kokoro TTS (82M parameters, runs on CPU, Apache 2.0) and Qwen3-TTS (10 languages, voice cloning from 3 seconds, Apache 2.0).
7. Cartesia Sonic — Best for Real-Time Voice Agents ($4/mo+)
If you were using Play.ht's API for voice bots or real-time applications, Cartesia Sonic 3 blows everything else away on latency. Time-to-first-audio of ~40ms on the Turbo variant — that's 4–5x faster than any competitor. Cartesia also offered migration assistance specifically to displaced Play.ht users during the transition period.
The Pro plan starts at $4/month with voice cloning from 3 seconds of audio. For detailed pricing across all tiers and comparison with other voice agent TTS options, check our Cartesia pricing guide and Cartesia review.
Play.ht Alternatives at a Glance
| Service | Price | Voice Cloning | Best For |
|---|---|---|---|
| ElevenLabs | $5–$99/mo | 30 sec, pro quality | Overall best quality |
| Fish Audio | $15/1M chars | Yes, 80+ languages | Best quality per dollar |
| Murf AI | Free 6mo, then $23/mo | Enterprise only | Easiest migration + free offer |
| Amazon Polly | $4–$16/1M chars | No | AWS developers, cheapest API |
| OpenAI TTS | $15/1M chars | No | OpenAI SDK users |
| Chatterbox | Free (MIT) | 5 sec, self-hosted | No vendor lock-in |
| Cartesia | $4/mo+ | 3 sec, instant | Real-time voice agents (40ms) |
How to Migrate from Play.ht (Step by Step)
Whether you're moving API integrations or just picking a new tool for occasional use, here's the practical migration path:
- Audit what you were using Play.ht for. API-first developers should look at Fish Audio, ElevenLabs, or Cartesia. Web app users should try Murf (especially with the free migration offer) or ElevenLabs. Open-source advocates should start with Chatterbox.
- Re-create voice clones. You'll need the original reference audio since Play.ht deleted all clone data. ElevenLabs needs 30 seconds, Fish Audio needs 10–30 seconds, Chatterbox needs 5 seconds, and Cartesia needs just 3 seconds. If you don't have the original audio, you'll need to re-record.
- Update API endpoints. Most modern TTS APIs use similar REST patterns. The main differences are in authentication, response format (streaming vs complete), and billing units (characters vs bytes vs seconds). Budget 2–4 hours for a typical API migration.
- Test at your actual usage level. Use our TTS cost calculator to estimate monthly costs before committing. Some services are cheap at low volume but expensive at scale (and vice versa).
- Consider hybrid approaches. Several former Play.ht power users I've spoken with now use ElevenLabs for high-quality voice work and an open-source model (Kokoro or Chatterbox) for bulk processing. This keeps costs under control while maintaining quality where it matters.
Cost Comparison at Real Usage Levels
Play.ht's pricing was $39/mo (Pro) or $99/mo (Business) with character limits. Here's what equivalent usage costs on each alternative:
| Service | 10K chars/mo | 100K chars/mo | 1M chars/mo |
|---|---|---|---|
| ElevenLabs | $5 (Starter) | $22 (Creator) | $99 (Pro) |
| Fish Audio | Free tier | ~$1.50 | $15 |
| Murf AI | Free (6mo offer) | $23 (annual) | $66+ (Business) |
| Amazon Polly | Free tier | Free tier | $4–$16 |
| OpenAI TTS | $0.15 | $1.50 | $15 |
| Chatterbox | $0 | $0 | $0 |
| Cartesia | $4 (Pro) | $4–$39 | $39–$239 |
For interactive cost estimates at your exact usage level, try our TTS cost calculator — it compares all 11 services we track. Our full TTS pricing comparison breaks down every plan, tier, and hidden cost.
What About "PlayAI"? Can I Still Use It?
No. After the acquisition, Meta's team became part of "Superintelligence Labs" — the Play.ht / PlayAI brand was retired entirely. People searching for "PlayAI text to speech" are finding nothing because the product doesn't exist as a consumer-facing service anymore. Meta is using the technology internally for AI Characters and wearable products, not as a standalone TTS service.
There's also no "Play.ht 2.0" or successor product coming. This was an acqui-hire — Meta wanted the team and the tech, not the business. If you see any site claiming to be "the new Play.ht," it's unrelated.
Can I Recover My Play.ht Data?
No. Meta deleted all user data in accordance with its internal privacy policies at shutdown. This includes:
- All generated audio files
- Voice clone models and reference audio
- Project files and scripts
- API keys and account data
- Billing history
If you downloaded your generated audio before the December 31 cutoff, you still have those files locally. But the voice clones, projects, and account settings are gone permanently. Multiple Reddit threads document users who lost months of work because no migration window was provided.
Which Alternative Should You Pick?
"I want the best quality, period"
→ ElevenLabs ($5/mo to start). Top-3 on every benchmark.
"I need to keep API costs low"
→ Fish Audio ($15/1M chars). Best blind-test results at the lowest price.
"I never want to get burned by a shutdown again"
→ Chatterbox (free, MIT). Self-hosted, open-source, can't be acquired away.
"I just need something working today with zero cost"
→ Murf AI (free 6-month migration offer). Claim the offer, evaluate others during those 6 months.
For a full comparison of every TTS service we track — including live voice demos you can listen to — check our best text-to-speech guide and best TTS API comparison.
Related Guides
By TextToLab Research Team · Last verified June 2026. Shutdown timeline from Bloomberg, Meta, and Audiostack. Pricing from official product pages. Murf migration offer confirmed via murf.ai. ElevenLabs and Murf affiliate links disclosed — all other recommendations are independent.