PlayHT alternatives in 2026 after the shutdown
PlayHT shut down on 31 December 2025, and its API, accounts and voice clones were deleted. Our guide compares seven replacement text to speech and voice cloning APIs, with 2026 pricing and a quick migration path. The picks run from a $0 self hosted model to premium narration, and PyAI takes over old PlayHT calls through an OpenAI style speech endpoint.

On this page
Why PlayHT users must switch to an alternative
PlayHT, later Play.ai, was a text to speech and voice cloning API with a large voice range and voice agent tooling. Meta acquired it, and the service was wound down, so the search is now a forced migration rather than a comparison of live plans.
The shutdown timeline
Meta acquired PlayHT in July 2025, the API shut down on 26 July 2025, and the platform was permanently terminated on 31 December 2025.
Deleted voice clones
Stored audio and voice clones were removed with no export, so any cloned voice has to be re-enrolled on a new provider.
No migration path
No official migration route was offered, so integrations that called the PlayHT API stopped and need re-pointing to a new one.
Cost at scale, historically
Before the shutdown, PlayHT held 4.2 out of 5 across 89 G2 reviews, with expensive plans and mispronunciation among the listed cons. PlayHT's old Professional $39 and Premium $99 plans are historical only.
PlayHT alternatives at a glance
Seven replacements, scored on price, cloning, commercial rights and job fit, using vendor pricing pages checked on 20 and 30 October 2026.
| Tool | Built for | Features | Free option | Price (2026) | Cloning |
|---|---|---|---|---|---|
| PyAI | Developers re-pointing a stranded PlayHT build | Speak, Cast, Hear and Omni under one key, OpenAI compatible speech, free cloning and prompt designed voices, WAV, PCM and G.711 | $50 credit, no credit cardSandbox keys never bill | $0.04/min Speak, $0.02/min Cast | Free |
| ElevenLabs | Ex PlayHT creators wanting expressive narration and dubbing | Narration, cloning, dubbing and transcription in one account, with API streaming | Free plan | $6 Starter, Flash API $0.04 per 1K chars | Yes |
| Cartesia | Builders who ran live agents, not batch voiceover | Sonic streaming models and a Line voice agent platform | 20,000 credits | $5 Pro, $49 Startup, $299 Scale | Yes |
| Fish Audio | Podcasters and YouTubers working on a small budget | A community voice library and open Fish Speech weights you can self host | Free plan, personal use only | Plus $11, Pro $75 a month billed yearly | Yes, on paid plans |
| Deepgram | Teams needing speech output and transcription on one meter | Aura text to speech beside Nova transcription and a Voice Agent API | $200 credit | Aura-2 $0.030 per 1K chars | No |
| Chatterbox | Developers wanting cloning no acquisition can switch off | Open weights under the MIT license, zero shot cloning, a multilingual variant, PerTh watermark | Free, MIT license | $0 plus your GPU | Yes, self hosted |
| Speechmatics | High volume IVR prompts and plain article narration | A text to speech API from a broad language transcription specialist | 1M characters a month | $0.011 per 1K chars | No |
| PlayHT (for comparison) | Creators left stranded by a discontinued API | A former text to speech and voice API, now shut down with no replacement | NoneShut down 31 Dec 2025 | No current pricing | Deleted |
How much cheaper PlayHT alternatives are
PlayHT sells nothing today, so the number that matters is what its replacements charge. Each bar converts a published rate into the bill for one million characters, roughly 1,100 minutes of audio, and the note says where that tool earns its place.
Replacement voice APIs, one million characters
PyAI Speak lands near $44 per million characters, with cloning at no extra cost
Bills for one million generated characters, from rates on each vendor page. Bar length is cost, and the note names the tradeoff.
- Speechmatics TTSPer character APIFirst 1M free
$11 per million characters on Pro, after one million free each month
- Fish AudioCreator cloningBudget cloning
Near $15 per million characters, estimated, not published, with cloning on Plus at $11 a month
- Deepgram Aura-2Developer TTSHigh volume
$30 a million characters on Aura-2, and new accounts open with $200 of credit
- PyAI SpeakReliable API, free cloningFree cloning
Roughly $44 a million characters at the Speak rate, cloning included. Cast at half the per minute price brings long narration near $22
- ElevenLabsPremium narrationNatural pacing in reviews
$40 on Flash, $80 on Multilingual per million characters, about five times the Fish Audio estimate and a voice quality reviewers rate well
- ChatterboxOpen sourceSelf hosted
No license fee. MIT weights run on your own GPU or Apple Silicon Mac, so the cost is hardware and setup hours
Rates checked 20 and 30 October 2026. PlayHT itself is discontinued, so no current rate is shown.
PlayHT alternatives reviewed
The seven picks cover the reasons ex PlayHT users move. PyAI re-points an API build with free cloning, ElevenLabs suits expressive narration, Cartesia suits low-latency streaming, Fish Audio and Chatterbox suit cheap cloning, Deepgram suits volume, and Speechmatics suits price.
PyAI
Built for re-pointing a dead PlayHT integration through an OpenAI style speech endpoint
PyAI is voice infrastructure, with four engines under a single API key: Speak for text to speech, Cast for long narration, Hear for transcription and Omni for phone agents. A stranded PlayHT build needs only a small code change, since Speak follows the OpenAI speech format, with 200 voices, free voice cloning and voices designed from a written prompt. Nothing in PyAI resembles a drag and drop studio, so creators who want a timeline editor will need a separate tool for that.
Per minute, Speak is $0.04 and Cast is $0.02. Cast is built for audiobooks and hour long videos. First audio arrives in about 53 ms warm in PyAI tests, against the 143 ms mean PlayHT published for Play 3.0. PyAI has no monthly plan, billing stops when generation stops, and an enrolled clone stays on the account with no expiry date.
With zero G2 reviews on record, PyAI cannot point to user ratings the way ElevenLabs can.
PyAI features
- Speak
- 200 stock voices behind an endpoint shaped like the OpenAI speech API, for app audio and voiceovers.
- Cast
- Hour long narration billed at two cents a minute and rendered in the background.
- Voice cloning
- Enrollment from a short clip costs nothing, and cloned speech bills at the normal Speak rate.
- Commercial use
- Covered by the standard Speak price, with no upgrade tier to unlock it.
PyAI pricing
Speak
$0.04/min
Voiceovers
Cast
$0.02/min
Long narration
Cloning
Free
Short clip
Metering is per minute, sandbox keys never bill, and a fresh account opens with $50 of credit and no credit card on file. Hear the voices on PyAI Speak.
PyAI pros and cons
Pros
- Speak follows the OpenAI speech request format, so a PlayHT integration that went dark on 26 July 2025 can be re-pointed with a small code change instead of a rebuild.
- Clone enrollment is free from a short clip, which matters because PlayHT deleted every stored clone on 31 December 2025 with no export.
- At $0.04 a minute, 200 minutes of speech a month costs about $8, and per minute metering means a quiet month bills close to nothing.
- First audio lands in about 53 ms warm and 98 ms cold in PyAI tests, versus a 143 ms mean PlayHT quoted for Play 3.0.
Cons
- No timeline or drag and drop editor, so podcast and voiceover creators who leaned on the PlayHT web app need a separate editing tool.
- PyAI launched recently, so it lacks the multi year record ElevenLabs or Deepgram bring to a procurement review after one vendor already vanished.
- Zero G2 reviews so far, and every latency number on this card comes from PyAI measurements in its US region, not an outside benchmark.
PyAI reviews
No user reviews of PyAI exist on G2 yet, so this tab stays honestly empty. Run your old PlayHT scripts through a free sandbox key, which carries $50 of credit and asks for no credit card, and judge the voices directly.
Choose PyAI if you are migrating a PlayHT API build and want a quick re-point, free cloning and a flat per minute rate.
ElevenLabs
Built for expressive English narration and dubbing for ex PlayHT creators
Most migration guides, including one ElevenLabs published on its own blog, send ex PlayHT creators here first. One account covers expressive narration, cloning, dubbing and transcription, with streaming through the API.
A free plan exists, and ElevenLabs' pricing page lists Starter at $6 and Pro at $99 a month. Through the API, Flash bills $0.04 and Multilingual $0.08 per 1,000 characters. G2 rates it 4.5 out of 5 across 1,260 reviews.
The tradeoff for a migrating PlayHT user is cost shape, since billing is per character and the richest voices sit on the dearer plans, so a high volume audiobook or ad run climbs faster than a flat per minute rate. ElevenLabs fits a creator whose PlayHT work was expressive English narration and who will absorb per character pricing for delivery, rather than a developer who mainly wants a cheap, like for like API swap.
ElevenLabs features
- Plans
- ElevenLabs lists Free, Starter at $6, Creator at $22, Pro at $99, Scale at $299 and Business at $990 a month on its pricing page checked 20 and 30 October 2026.
- Flash API
- The Flash model bills $0.04 per 1,000 characters through the API and runs about 135 ms end to end by published figures.
- Multilingual API
- Multilingual output costs $0.08 per 1,000 characters, double the Flash rate, for narration outside English.
- Cloning
- Voice cloning sits on the same account as narration, so a clone lost in the PlayHT shutdown can be enrolled again here.
- Dubbing and transcription
- Dubbing and speech to text share the ElevenLabs account, which saves a creator from opening several vendor accounts.
- Review base
- 1,260 G2 reviews average 4.5 and Capterra scores 4.7, praising voice quality, speed and the size of the voice selection.
Against PlayHT's pain points: ElevenLabs
Several migration guides, including ElevenLabs' own, point ex PlayHT creators here, where a large review base meets a per character premium.
- Vendor risk: ElevenLabs carries 1,260 G2 reviews, a deep record next to newer vendors, though ownership stability is not something the sources measure
- Clone rebuild: Cloning is built in, so a voice deleted when PlayHT closed on 31 December 2025 can be rebuilt from old source recordings
- API switch: Streaming runs through the ElevenLabs API, so code that once called PlayHT can be re-pointed to Flash or Multilingual models
- Scale cost: A million characters costs $40 on Flash and $80 on Multilingual, and plans jump from $99 Pro to $299 Scale as a catalog grows
- Pronunciation: G2 cons still cite pronunciation slips, and one reviewer says the model stumbles over legal citations, the same gripe PlayHT drew
- Support: ElevenLabs handles support by email, with faster response reserved for higher plans, so confirm what the tier you pick includes before you rely on it
ElevenLabs pros and cons
Pros
- G2 users score it 4.5 out of 5 over 1,260 reviews and Capterra 4.7 from 25 reviews, praising voice quality, speed and the size of the voice selection.
- Cloning, dubbing and speech to text share one account, so a creator rebuilding a PlayHT podcast workflow does not juggle several vendors.
- Flash runs about 135 ms end to end by published figures, under the 143 ms mean PlayHT gave for Play 3.0.
Cons
- Capterra reviewers single out monthly credits that run out fast, which bites when a backlog of PlayHT audiobooks has to be regenerated.
- Multilingual at $0.08 per 1,000 characters comes to about $80 a million, over five times the Fish Audio estimate.
- G2 cons cite pricing complaints and pronunciation slips, the same two gripes reviewers raised about PlayHT itself.
- Plans step from $6 Starter to $99 Pro and $299 Scale, so a growing catalog can force a big jump in monthly spend.
ElevenLabs reviews
ElevenLabs holds 4.5 on G2 from 1,260 reviews. ElevenLabs on G2 ↗
A reviewer highlights: "The absolute best part is just how genuinely human the voices sound: the natural pauses, the breathing, and the way emphasis shifts." (Pradumya K., Enterprise, G2 Oct 2026)
Another flags: "The main pain point is how it handles dense legal terminology, citations and Latin maxims. The model frequently stumbles over citations." (Shreyansh P., Small-Business, G2 Oct 2026)
On pricing: "The best part about ElevenLabs is the free credit thing they have." (Rashi R., Small-Business, G2 Oct 2026)
Choose ElevenLabs if you make PlayHT style audiobooks and ads that need a highly expressive English voice and can absorb a per character premium.
Cartesia
Built for streaming speech under live voice agents that once ran on Play 3.0
Cartesia's Sonic models stream speech in under about 100 ms, and it ships a Line voice agent platform, so it suits ex PlayHT builders who ran live agents more than batch voiceover.
Monthly tiers begin with 20,000 free credits, then Pro at $5, Startup at $49 and Scale at $299. Managed agents bill $0.03 a minute, plus $0.014 a minute when the call uses a Cartesia number.
The limit for an ex PlayHT creator is that Cartesia is an API and an agent runtime, not a voiceover studio, so there is no editor, media library or export workflow for a non technical user. Cartesia suits a team that ran live PlayHT agents and needs streaming speech that starts in about a tenth of a second, and that is comfortable building the voice into its own application rather than clicking through a timeline.
Cartesia features
- Sonic streaming
- Cartesia's Sonic models stream first audio in 82 to 100 ms by Cartesia figures, network included.
- Line agents
- The Line voice agent platform bills managed agents at $0.03 a minute, plus $0.014 a minute on a Cartesia phone number.
- Tiers
- 20,000 free credits start the ladder, then Pro at $5, Startup at $49 and Scale at $299 a month.
- Instant cloning
- Instant voice cloning unlocks on Pro and above, per the Cartesia pricing page checked 20 and 30 October 2026.
- Free trial
- A G2 reviewer notes that Cartesia offers a free trial, so a rebuilt PlayHT agent can be tested before any plan.
Against PlayHT's pain points: Cartesia
Cartesia fits ex PlayHT builders who ran live agents on Play 3.0 and now need fast streaming speech.
- Vendor risk: Only 43 G2 reviews exist so far, a thin track record for a buyer who has already lost one voice vendor
- Clone rebuild: Instant cloning starts on the $5 Pro tier, so a deleted PlayHT clone can be re-enrolled for little money
- API switch: Sonic streams through an API built for live agents, the job Play 3.0 did at a 143 ms mean by PlayHT's own figure
- Scale cost: Scale at $299 buys 8 million credits, about $0.037 per 1,000 characters, but agent calls add a per minute meter on top
- Pronunciation: Pronunciation quality is not covered in the Cartesia reviews checked, so script tests with brand names and numbers
- Support: A Small-Business G2 reviewer wants deeper per call reporting as agents grow. Support terms themselves are not published
Cartesia pros and cons
Pros
- Sonic streams first audio in 82 to 100 ms by Cartesia figures, network included, quick enough to sit under a live agent that used Play 3.0.
- Pro at $5 a month lets a team rebuild a PlayHT agent prototype before committing to Startup at $49.
Cons
- Every latency figure is vendor published, so time your own scripts before trusting them on production calls.
- Agent calls carry two meters, $0.03 a minute of agent balance plus $0.014 a minute on a Cartesia phone number, which makes a monthly bill harder to forecast.
- Only 43 G2 reviews so far, a thin record for a team that has already been burned by one vendor exit.
Cartesia reviews
Cartesia holds 4.3 on G2 from 43 reviews. Cartesia on G2 ↗
A reviewer highlights: "The biggest advantage is the low latency, which makes conversations feel more natural and responsive, and the voice quality is quite natural, especially for conversational use cases." (Yogita B., Mid-Market, G2 Sep 2026)
Another flags: "Some areas could provide more detailed information around individual calls. I would also like more flexibility and deeper reporting as the number of agents and calls grows." (Alison D., Small-Business, G2 Sep 2026)
On pricing: "The best part is Cartesia provides a free trial too, so I try such free alternatives before purchasing any plan." (Verified user, Small-Business, G2 Oct 2026)
Choose Cartesia if you ran live PlayHT agents and need streaming speech that starts in about a tenth of a second.
Fish Audio
Built for cheap cloned voices for podcasters and YouTubers on a small budget
Fish Audio costs less for hobby and creator work, with a community voice library and open Fish Speech weights for anyone who wants to host the model.
The free tier is limited to personal projects. Plus at $11 a month unlocks commercial use, Pro is $75 and Max $749, and guides put a million characters near $15.
The catch is scope and support, since the free tier is personal use only, the catalog is community sourced rather than curated, and help comes through a community channel rather than an account team. Fish Audio fits a YouTuber or podcaster migrating from PlayHT who wants commercial cloning at a low monthly rate, or the open Fish Speech weights to self host, not an enterprise that needs guaranteed support and uptime.
Fish Audio features
- Plus plan
- Fish Audio Plus costs $15 a month, or $11 billed yearly, and unlocks commercial use of cloned voices.
- Higher tiers
- Pro is $100 a month or $75 billed yearly, and Max is listed at $749 on G2.
- Open weights
- The Fish Speech weights can run on a creator's own GPU, though the license is non commercial, so commercial use still needs the Plus plan at $11 a month.
- s2-pro API
- The s2-pro API bills $15 per million UTF-8 bytes, about $0.015 per 1,000 characters, per the Fish docs.
Against PlayHT's pain points: Fish Audio
Fish Audio is the budget pick for creators replacing a lost PlayHT clone, with an open model as a fallback.
- Vendor risk: Open Fish Speech weights mean the model survives on your own disk even if the hosted service ever changes hands
- Clone rebuild: Cloning comes on every plan, and commercial rights for the cloned voice start on Plus at $11 a month billed yearly
- API switch: An s2-pro API exists for developers re-pointing a PlayHT integration, serving creators and developers alike
- Scale cost: The API costs about $15 per million characters, roughly a fifth of the ElevenLabs Multilingual rate
- Pronunciation: A G2 reviewer reports regenerating lines because one word comes out wrong, a cost in time rather than money
- Support: Support terms are not published in the Fish Audio sources checked, and no Capterra reviews exist to cross check
Fish Audio pros and cons
Pros
- Plus at $11 a month buys commercial cloning, about a ninth of the ElevenLabs Pro price, for a podcaster replacing a deleted PlayHT voice.
- Open Fish Speech weights let a creator move the same model onto a home GPU if the hosted service ever changes owners.
- G2 reviewers give it 4.5 out of 5 across 119 reviews, level with the ElevenLabs average.
Cons
- Commercial rights start on Plus, so a monetised channel that ran on the free PlayHT tier now pays $11 a month at minimum.
- Fish Audio has a narrower voice range than ElevenLabs, and no Capterra reviews exist to cross check the G2 score.
Fish Audio reviews
Fish Audio holds 4.5 on G2 from 119 reviews. Fish Audio on G2 ↗
On pricing: "I like the smoothness and speed of generation, the quality of the audio, and the very cheap price, especially with the S2 model." (Chiagozie O., Small-Business, G2 Jul 2026)
Choose Fish Audio if you are a YouTuber or podcaster who needs a cloned voice cleared for commercial use at $11 a month.
Deepgram
Built for teams that need speech output and transcription on one usage meter
Deepgram is the developer voice API ex PlayHT teams pick for scale, pairing Aura-2 text to speech with Nova-3 transcription and a Voice Agent API under one account.
Aura-2 bills $0.030 per 1,000 characters, the Voice Agent API is $0.075 a minute, and a new account receives $200 of credit that never expires.
The reservation is that Deepgram grew up in transcription and Aura-2 is the younger text to speech side, so its voice range is narrower than a studio built for narration. Deepgram works for an ex PlayHT engineering team generating millions of characters a month that also transcribes calls and wants speech, transcription and agents on one invoice, rather than a creator who wants an editing timeline and a deep catalog of character voices.
Deepgram features
- Aura-2 voices
- Deepgram's Aura-2 text to speech bills $0.030 per 1,000 characters.
- Nova-3 transcription
- Nova-3 transcription shares the same usage meter as Aura-2, so a voice app that also listens needs one contract.
- Voice Agent API
- A Voice Agent API at $0.075 a minute bundles listening, reasoning and speaking for teams rebuilding PlayHT agents.
- Non expiring credit
- New accounts receive $200 of credit that never expires, about 6.6 million Aura-2 characters.
- Review base
- Deepgram holds 4.6 out of 5 from 482 G2 reviews, with accuracy, speed and live transcription praised most.
Against PlayHT's pain points: Deepgram
Deepgram suits ex PlayHT apps that generate speech at volume and also transcribe, though it offers no cloning.
- Vendor risk: 482 G2 reviews give Deepgram a long public record, while its ownership outlook is not covered in the sources
- Clone rebuild: Cloning is missing entirely, so a product built on a PlayHT clone has to add a second vendor for that voice
- API switch: Aura-2 and the Voice Agent API are developer endpoints, the natural target for code that called the PlayHT API
- Scale cost: At $0.030 per 1,000 characters, a million characters costs $30, and the $200 credit regenerates a large archive first
- Pronunciation: Aura-2 pronunciation is not rated in the sources checked. The jargon complaint in Deepgram reviews concerns transcription
- Support: Support terms are not published on the Deepgram pricing checked, and there is no studio for creators who used a web editor
Deepgram pros and cons
Pros
- G2 users rate Deepgram 4.6 out of 5 across 482 reviews, citing accuracy, speed and real time transcription.
- Text to speech and Nova-3 transcription sit on one usage bill, so a PlayHT voice app that also listens needs a single contract.
- The $200 starting credit covers about 6.6 million Aura-2 characters, enough to regenerate a large PlayHT archive before paying.
- The Voice Agent API at $0.075 a minute packages listening, reasoning and speaking for teams rebuilding PlayHT agents.
Cons
- Cloning is missing entirely, so a product that depended on a PlayHT clone must pair Deepgram with a second vendor.
- G2 reviewers mention limited language support and pricing complaints, and there is no studio for creators who worked in a web editor.
Deepgram reviews
Deepgram holds 4.6 on G2 from 482 reviews. Deepgram on G2 ↗
A reviewer highlights: "Deepgram is really a developer-friendly solution and is extremely fast if you compare it against other vendors like OpenAI Whisper and Speechmatics." (Diego P., Enterprise, G2 Aug 2026)
Another flags: "Although the accuracy is very good, it can sometimes be a bit difficult to adjust the models for very specific cases or when there is technical jargon." (Verified user, Small-Business, G2 Aug 2026)
On pricing: "It supports word recognition and speaker diarization, and it also has a very good price offering comparatively." (Akshay M., Mid-Market, G2 Aug 2026)
Choose Deepgram if you run apps generating millions of characters a month that also transcribe calls and want both on one invoice.
Chatterbox
Built for MIT licensed cloning that no acquisition can switch off
Resemble AI publishes Chatterbox as open weights under the MIT license, with zero shot cloning from roughly 5 to 10 seconds of reference audio and a Multilingual variant. Output carries a PerTh watermark.
Running it locally costs hardware and an afternoon of setup, and DeepInfra hosts it at $10 per million characters for teams without a GPU. Resemble AI itself holds 3.9 out of 5 on G2 from 20 reviews.
The cost here is effort rather than money, since running Chatterbox means owning a GPU, the setup and any tuning, and every output carries a PerTh watermark that some uses will not want. Chatterbox is the pick for a developer burned by the PlayHT shutdown who wants cloning with no per minute fee and no vendor that can disappear, and who can either self host or lean on a budget host, rather than a team that wants a managed studio out of the box.
Chatterbox features
- MIT license
- Resemble AI publishes Chatterbox as open weights under MIT, allowing commercial use with no per character fee.
- Zero shot cloning
- Roughly 5 to 10 seconds of reference audio is enough to clone a voice, and a Multilingual variant exists.
- PerTh watermark
- Every Chatterbox clip carries a PerTh watermark, which matters for client voiceover work.
- Hosted option
- DeepInfra hosts Chatterbox at $10 per million characters for teams that have no GPU of their own.
Against PlayHT's pain points: Chatterbox
Chatterbox is the answer to the PlayHT lesson for developers: weights on your own disk that no acquisition can delete.
- Vendor risk: Weights stored locally under MIT cannot be switched off or deleted the way PlayHT voice clones were
- Clone rebuild: A 5 to 10 second sample is enough, so old recordings of a lost PlayHT voice can seed a new clone
- API switch: The open model ships with no hosted API of its own, so re-pointing code means self hosting or vetting DeepInfra
- Scale cost: Local runs cost hardware and setup only, DeepInfra charges $10 per million characters, and concurrent streams become your capacity problem
- Pronunciation: Pronunciation quality is not reviewed in the sources checked, and Resemble AI holds 3.9 on G2 from 20 reviews
- Support: No support desk comes with the open weights, so fixes depend on your own engineers or the hosting vendor
Chatterbox pros and cons
Pros
- MIT licensing allows commercial use with no per character fee, and weights on your own disk cannot be deleted the way PlayHT clones were.
- A 5 to 10 second sample is enough to clone, so re-enrolling a lost PlayHT voice from old recordings is realistic.
Cons
- A capable GPU or an Apple Silicon Mac plus setup time is required, which a creator without engineering help may not have.
- The open model ships with no hosted API or support desk, and DeepInfra at $10 per million characters is a separate vendor to vet.
- Every clip carries a PerTh watermark, worth knowing before shipping client voiceovers.
- Scaling to many concurrent streams becomes your own capacity problem, unlike a metered API that absorbs load spikes.
Choose Chatterbox if you want cloning with no per minute cost or vendor risk.
Speechmatics
Built for a per character TTS bill near $11 for a million characters
Speechmatics built its name on transcription, scoring 4.7 out of 5 across 75 G2 reviews, and added a per character TTS API more recently.
Pro tier TTS bills about $0.011 per 1,000 characters, and the free tier includes one million characters a month, roughly 20 hours of English.
The tradeoff is expression, since Speechmatics came from transcription and its text to speech is plain and functional rather than performance grade, so emotional reads and character voices are not its strength. Speechmatics fits a high volume PlayHT replacement where the cost per thousand characters and a generous free tier matter more than expressive delivery, such as bulk IVR prompts or straightforward article narration.
Speechmatics features
- Pro TTS rate
- Speechmatics text to speech bills about $0.011 per 1,000 characters on the Pro tier.
- Free characters
- The free tier includes one million characters a month, roughly 20 hours of English audio.
- Transcription heritage
- Speechmatics built its name on transcription and scores 4.7 out of 5 across 75 G2 reviews.
- Two directions, one vendor
- Transcription and synthesis come from the same vendor, whose G2 reviewers praise accuracy and efficiency.
- Plain narration
- Voices sound plainer than ElevenLabs, which shows less on phone prompts and alerts than on expressive reads.
- Responsive support
- A Small-Business G2 reviewer calls Speechmatics customer support reliable with quick responses.
Against PlayHT's pain points: Speechmatics
Speechmatics is the low cost per character option for plain PlayHT narration, without any cloning.
- Vendor risk: Speechmatics has a long transcription record, though text to speech is its newer product line
- Clone rebuild: Cloning is absent, so a deleted PlayHT clone cannot be rebuilt on Speechmatics at any price
- API switch: A per character TTS API takes over plain PlayHT narration calls, with one million free characters to test the switch
- Scale cost: Pro bills near $11 per million characters, below the $15 Fish Audio API figure and just above DeepInfra's $10 for Chatterbox
- Pronunciation: Pronunciation is not rated in the sources checked, and G2 reviewers flag limited features on the newer TTS line
- Support: G2 praise for reliable, quick support is the clearest support signal among the six other tools
Speechmatics pros and cons
Pros
- One million free characters every month, about 20 hours of English, covers testing a whole PlayHT catalog before any bill.
- Near $11 per million characters on Pro, a quarter of the PyAI Speak cost per character for plain narration.
- Transcription and synthesis come from one vendor whose G2 reviewers praise accuracy and efficiency.
Cons
- Cloning is absent, so a deleted PlayHT clone cannot be rebuilt on Speechmatics.
- TTS is the newer product line, and G2 reviewers flag limited features and limited language support.
- Voices sound plainer than ElevenLabs, a tradeoff that shows on expressive narration more than on phone prompts or alerts.
Speechmatics reviews
Speechmatics holds 4.7 on G2 from 75 reviews. Speechmatics on G2 ↗
A reviewer highlights: "We appreciate the relative ease of integrating Speechmatics into our broadcast monitoring workflow. Customer support is reliable and provides quick responses." (Joe T., Small-Business, G2 Aug 2025)
Another flags: "It appears to fill in a whole lot of nonsense when it hears noise or background chatter, creating disconcerting false positive transcriptions." (Verified user, Mid-Market, G2 Sep 2026)
On pricing: "Functionalities and pricing tailored to your exact application and business domain, a good software stack and API, reliable transcoding." (Verified user in broadcast media, Enterprise, G2 Feb 2023)
Choose Speechmatics if you need high volume plain narration where a cent per thousand characters matters more than expressive delivery.
PlayHT alternatives by job and use case
Match the replacement to the part of PlayHT you lost.
Replacement picks by job
| Job | Pick | Why |
|---|---|---|
| Re-point an API build | PyAI Speak | OpenAI compatible, $0.04 a minute |
| Audiobooks and long narration | PyAI Cast, ElevenLabs | Cast at two cents a minute, ElevenLabs for expressive reads |
| Voice agents | Cartesia, PyAI Omni | Streaming agents, Omni at $0.05/min + $0.01 telephony |
| Large developer TTS volume | Deepgram | Aura-2 and transcription on one bill |
| Self hosted cloning | Chatterbox, the flow-matching voice-cloning model | Free, your GPU, no shutdown risk |
| Low cost per character | Speechmatics | $0.011 per 1K, 1M free monthly |
Text to speech by price
| Tool | Price |
|---|---|
| Speechmatics TTS | $11 per 1M characters |
| Fish Audio | Near $15 per 1M characters |
| PyAI Cast | $0.02/min, about $22 per 1M characters |
| Deepgram Aura-2 | $30 per 1M characters |
| ElevenLabs Flash | $0.04 per 1K characters |
Open source voice cloning
| Model | License | Note |
|---|---|---|
| Chatterbox | MIT | Multilingual variant, clones from 5 to 10 seconds |
| the flow-matching voice-cloning model | Open | Flow matching, zero shot cloning, self hosted |
Free and open source PlayHT alternatives
Open models remove the risk that stranded PlayHT users. Chatterbox under the MIT license and the flow-matching voice-cloning model both clone voices locally, trading per minute fees and vendor exposure for a GPU and setup hours. On the hosted side, trial credit keeps side by side tests cheap: PyAI gives $50 with no credit card, Cartesia 20,000 credits, Deepgram $200, and Speechmatics one million characters each month.
When owning your voices makes sense
The lesson from the PlayHT shutdown is to weigh ownership and portability, not just price. A standard, OpenAI compatible API means a future move costs a config change, not a rewrite. PyAI voices can be auditioned by language and accent in the PyAI voice directory. Keeping an export of prompts and source audio, and favouring a self hostable open model for critical workloads, both reduce the risk of being stranded again.
How to migrate off PlayHT in 4 steps
Four steps cover an API build, and re-enrolling clones plus regenerating audio take the longest.
Step 1 of 4
Pick a replacement API
Shortlist a text to speech API whose request format matches the OpenAI speech endpoint, which keeps the code diff small.
Step 2 of 4
Re-enroll your voices
Record a clean 10 to 30 second sample and re-enroll any voices PlayHT deleted.
Step 3 of 4
Regenerate critical audio
Rebuild the audio you still need, and test your longest scripts before going live.
Step 4 of 4
Switch and keep an export
Move traffic to the new API and keep an export of prompts and audio for portability.
Which PlayHT alternative to pick
With PlayHT gone, the choice turns on what you rebuild first. ElevenLabs suits expressive English, Cartesia live agents, Deepgram large volumes with transcription, and Chatterbox anyone who wants weights on their own hardware. PyAI Speak covers the API re-point at $0.04 a minute with cloning included, and a sandbox key comes with $50 of credit and no credit card required.
More text to speech comparisons
PlayHT alternatives FAQs
Common questions from teams migrating off PlayHT.
Is PlayHT still available?
No, it is shut down. Meta acquired PlayHT in July 2025, PlayHT switched off its API on 26 July 2025 and closed the whole service for good on 31 December 2025. Accounts, stored audio, voice clones and API integrations were deleted, and no migration path was provided, so users have to move to another voice API.
What is a good PlayHT alternative?
Match it to the job. ElevenLabs suits expressive English narration, Cartesia streaming agents, and Fish Audio and Speechmatics tight per character budgets. Deepgram fits apps that also transcribe, and Chatterbox runs free on your own GPU. PyAI Speak fits teams wanting free cloning, a flat $0.04 per minute rate and an OpenAI compatible API for old PlayHT calls.
What replaces the PlayHT API?
An OpenAI compatible text to speech API. PyAI Speak uses the OpenAI speech format, so pointing an old PlayHT integration at it needs little rework, at $0.04 a minute. ElevenLabs, Deepgram and Cartesia also offer streaming text to speech APIs, each with its own format to adapt to.
Can I move my PlayHT voice clones?
No, they were deleted. PlayHT removed stored voice clones when it shut down, with no export, so clones have to be re-enrolled. PyAI enrolls a cloned voice for free from a short clip, ElevenLabs and Resemble AI offer hosted cloning across languages, and Chatterbox clones open source on your own hardware.
What happened to Play.ai and PlayAI?
Same product, same shutdown. PlayHT rebranded to Play.ai, also seen as PlayAI, before Meta acquired it and wound it down on the same timeline. Searches for any of the three names point to the discontinued service, so the replacement options on this page apply to all of them.
How do I migrate off PlayHT fast?
Pick an API, re-enroll, regenerate. Choose a replacement text to speech API, re-enroll any cloned voices, regenerate the audio you still need, and test your longest scripts before going live. PyAI gives $50 of credit with no credit card, and its OpenAI compatible format keeps the code change small.
Test PyAI on your own audio and calls.
Get a sandbox key and hear the output in under a minute. Sandbox keys are free, with $50 in starter credit and no credit card required. Speak is $0.04/min, Cast $0.02/min, Hear $0.001/min and Omni $0.05/min + $0.01 telephony.