Google Speech to Text alternatives in 2026 for STT and TTS
Seven speech to text and text to speech APIs that replace Google Cloud Speech without a GCP project, weighed on per minute rates, signup credit and cloning. AssemblyAI and Deepgram are transcription APIs that G2 reviewers rate 4.6, ElevenLabs is known for expressive voices, and PyAI pairs $0.001 a minute transcription with free cloning.

On this page
Why developers look for Google Cloud Speech alternatives
Google Cloud Speech is a capable pair of APIs, speech to text and text to speech, with wide language coverage. Teams look elsewhere over price, setup and accuracy, not its reach.
Transcription price
Speech to Text lists $0.016 a minute on V2 for the first 500,000 minutes each month and reaches $0.004 only past 2 million, so low and mid volume pays the top rate. G2 reviewers name expense and pricing as their main complaint.
Google Cloud setup
Using the APIs means a Google Cloud project, billing, IAM roles and service accounts, which is heavy for a team that just wants speech.
Accuracy on tough audio
G2 reviewers of Google Cloud Speech report accuracy issues that need manual correction on noisy recordings, so phone and accented audio deserve a side by side test against specialist engines.
No free cloning
Google offers a paid Instant Custom Voice at $60 per million characters, not a free self serve clone, so a branded voice costs extra.
Google Cloud Speech alternatives at a glance
Seven replacements, set side by side on transcription rate, voice output, signup credit and cloning, using vendor pricing pages read on 20 and 21 October 2026.
| Tool | Built for | Features | Free option | Price (2026) | Cloning |
|---|---|---|---|---|---|
| PyAI | Teams that want speech to text and voices on one bill | Hear transcription, Speak and Cast voices, the Omni speech to speech model, an OpenAI compatible API, and WAV, PCM and G.711 output | $50 credit, no credit cardSandbox keys never bill | Hear $0.001/min, Speak $0.04/min | Free |
| ElevenLabs | Studios producing expressive narration | A large voice library, instant voice cloning, dubbing, and a studio workspace for long form audio | 10,000 credits a month | Starter $6, Creator $22 a month | Instant cloning from Starter |
| Cartesia | Agent builders needing streaming voice | The Sonic model for low latency streaming speech and an on device option for developers | 20,000 credits | Pro $5, Startup $49, Scale $299 | Not stated on pricing page |
| AssemblyAI | Teams that put transcription accuracy first | Accurate transcription with speaker labels, summaries, and audio intelligence models run over the audio | $50 credit | Universal-2 about $0.0025/min | No |
| Deepgram | Developers wiring speech into apps | Nova transcription, Aura voices, and a single voice agent endpoint built for developer apps | $200 credit | Nova-3 $0.0077/min streaming, Aura-2 $0.030/1K chars | No |
| Azure Speech | Teams standardizing on a second cloud | Cloud speech to text, neural text to speech, and a custom neural voice add on | 5 STT hours a month | Per hour STT, per character TTS | Custom voice add on |
| Speechmatics | Teams transcribing many accents | Accent and dialect aware transcription, real time and batch, with on premise deployment | $100 credit | Melia-1 $0.24/hr batch | No |
| Google Cloud Speech (for comparison) | Teams already building on Google Cloud | Speech to text and text to speech on GCP, with a paid custom voice build | 60 STT minutes a monthGCP project needed | STT from $0.016/min, TTS $4 to $160/1M | Paid custom voice |
Google Cloud Speech to Text pricing vs alternatives
Google Cloud bills speech to text per minute and text to speech per character, so the clearest comparison is the transcription rate, where the gap is widest. The chart prices speech to text per minute. For text to speech, Google is cheap at $4 per million on Standard but $160 on Studio, which specialist voices beat on quality.
Speech to text APIs, cost per minute
Per minute transcription rates against Google Speech to Text V2
Pay as you go rates from each vendor, converted to a per minute figure where the vendor quotes per hour. A longer bar means a higher bill.
- PyAI HearSpeech to text200 ms partials
$0.001 per audio minute, with 4.76% word error on 8 kHz phone audio in PyAI tests
- AssemblyAIUniversal-2 batchAudio intelligence
About $0.0025 a minute, from $0.15 an hour, with $50 of signup credit
- SpeechmaticsMelia-1 batchAccent coverage
About $0.004 a minute at $0.24 an hour, plus $100 of free credit
- Deepgram Nova-3StreamingLow latency
$0.0077 a minute streaming on pay as you go, $0.0043 pre recorded, with $200 of credit
- Google Cloud STTThe tool comparedGCP setup
$0.016 per minute on V2 for the first 500,000 minutes, falling to $0.004 past 2 million
Rates checked 20 and 21 October 2026 on each vendor pricing page. Deepgram rates rechecked 30 October 2026.
Google Cloud Speech competitors reviewed
The seven picks cover both sides of Google Cloud Speech. PyAI pairs cheap transcription with free cloning, Deepgram and AssemblyAI are speech to text APIs that G2 reviewers rate 4.6, while ElevenLabs and Cartesia are rated for natural speech, Azure is the other big cloud, and Speechmatics on languages.
PyAI
Built for transcription at $0.001 a minute plus voice output on one key, outside Google Cloud
PyAI is voice infrastructure for teams leaving Google Cloud Speech: Hear handles transcription, Speak and Cast handle voice output, and Omni runs phone agents, all on one API key with no Google Cloud project, IAM role or service account. Requests follow an OpenAI compatible format, so most code changes come down to an endpoint and a key. PyAI is a voice API rather than a full cloud, so a team using other Google Cloud services keeps those where they are.
Hear lists $0.001 a minute, Speak $0.04 a minute and Cast $0.02 for long narration. In PyAI tests, Hear posts 4.76% word error on 8 kHz phone audio and 1.59% on clean audio, with about 200 ms to the first partial transcript.
Outside reviews of Hear do not exist yet: PyAI has no G2 listing, so the figures above come from its own benchmark runs.
PyAI features
- Hear
- Streaming and batch speech to text at $0.001 per minute, with batch jobs running 92 to 247 times faster than real time.
- Speak
- Text to speech with 200 stock voices and first audio in about 53 ms when warm, returned as WAV, PCM or G.711.
- Voice cloning
- Enrolling a clone from a short clip costs nothing, and speech in that voice then bills at the normal Speak rate.
- Commercial use
- Commercial rights come with the standard per minute rate, with no upgrade to a business tier.
PyAI pricing
Hear
$0.001/min
Transcription
Speak
$0.04/min
Voiceovers
Cloning
Free
Short clip
Usage is metered by the minute, sandbox keys are never charged, and each new account opens with $50 of free credit and no credit card. Listen to the stock voices on PyAI Speak.
PyAI pros and cons
Pros
- Hear lists $0.001 a minute against $0.016 on Google V2, so 10,000 minutes a month costs about $10 instead of $160, with no GCP project to configure.
- Transcription and voice output share one key and one bill, where Google splits Speech to Text and Text to Speech into separate APIs.
- Voice cloning is free to enroll from a short clip, while Google bills Instant Custom Voice at $60 per million characters.
- PyAI has no monthly plan, so a quiet month on a side project costs only the minutes used.
Cons
- PyAI is only a voice API, so teams relying on other Google Cloud services still keep a GCP account for those.
- PyAI is a newer vendor than Google, with a shorter production record for a buyer to check.
- PyAI has no G2 reviews yet, and the 4.76% word error and 200 ms partial figures come from PyAI tests, not outside benchmarks.
PyAI reviews
Nobody has posted a G2 review of PyAI Hear or Speak yet, so this tab stays empty rather than inventing praise. A free sandbox key with $50 of credit and no credit card is enough to run your own recordings through Hear beside Google.
Choose PyAI if you want cheap transcription and speech under one key, without a Google Cloud project.
ElevenLabs
Built for expressive narration when Google Studio or Chirp 3 HD voices sound flat
ElevenLabs is the voice output pick when Google Studio or Chirp 3 HD voices sound too flat for audiobooks, ads or character reads. ElevenLabs sells expressive voices, instant cloning from the Starter plan and professional cloning from Creator, and it also runs a dubbing product.
The free plan gives 10,000 credits a month for non commercial use, then Starter costs $6 and Creator $22 a month, with Scale at $299 and Business at $990. Credits map to characters, so heavy narration burns through a plan quickly. G2 rates it 4.5 out of 5 across 1,260 reviews.
ElevenLabs is a text to speech pick first, though it also ships its own speech to text, so a team leaving Google usually comes to it for voice output rather than transcription. The paid tiers climb steeply from the $6 Starter to the $990 Business plan, so narration volume decides the cost. ElevenLabs fits narration, ads and character work where Google Studio or Chirp voices sound too plain to carry long reads.
ElevenLabs features
- Expressive voices
- ElevenLabs sells narration voices aimed at audiobooks, ads and character reads.
- Cloning tiers
- ElevenLabs unlocks instant clones at $6 a month on Starter, and professional clones need the $22 Creator tier.
- Plan ladder
- ElevenLabs runs Free, Starter $6, Creator $22, Scale $299 and Business $990 a month.
- Dubbing
- ElevenLabs includes a dubbing product in the same account as its voices.
- Own transcription
- ElevenLabs also offers speech to text, so voice output and transcripts share one login.
Against Google Speech to Text's pain points: ElevenLabs
ElevenLabs replaces the Google voice side, not a per minute Google transcription bill.
- Transcription price: ElevenLabs bills credits per character on a monthly plan, so no per minute rate sits next to the Google $0.016 V2 figure for a direct check
- Cloud setup: ElevenLabs signup is a web account with a monthly plan, with no cloud project, IAM roles or service accounts to configure
- Tough audio: ElevenLabs does not publish transcription accuracy on noisy audio in the sources checked, so run your own phone recordings through a trial
- Voice cloning: ElevenLabs starts self serve cloning at a $6 monthly plan, while a Google custom voice is billed at $60 for every million characters
- One vendor: ElevenLabs covers voices, dubbing and speech to text under one account, where Google splits speech into two APIs
- Free credit: ElevenLabs Free allows 10,000 monthly credits that cannot be used commercially and expire at month end
ElevenLabs pros and cons
Pros
- ElevenLabs voices sound expressive, and G2 reviewers rate the quality well, a step up from the Google Standard and WaveNet tiers at $4 per million characters.
- Instant cloning opens on the $6 Starter plan and professional cloning on the $22 Creator plan, with commercial rights from Starter.
- One account covers text to speech, dubbing and its own speech to text, so a creator team can stay with a single vendor.
Cons
- Credit plans drain fast at volume: G2 reviewers list expense and pricing as main complaints, and unused Free credits never roll over.
- Billing is per character on a monthly plan, so buyers coming from per minute Google transcription get no direct rate to compare.
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 need narration, ads and character voices where Google voices sound too plain.
Cartesia
Built for streaming voice output for live agents
Cartesia sells Sonic, a streaming text to speech model tuned for low latency, plus a Line platform for voice agents. Cartesia fits teams replacing Google Text to Speech inside a live agent rather than a batch voiceover job.
Plans run on credits, about one per character: Free gives 20,000, Pro costs $5, Startup $49 and Scale $299 a month for 8 million credits. Voice agents bill $0.03 a minute plus $0.014 on a Cartesia phone number, with LLM tokens charged separately.
Cartesia covers voice output only, so a team leaving Google still needs a separate transcription vendor, and LLM token charges make a talkative agent cost more than the per minute sticker. Cartesia fits teams replacing Google speech in a live agent that can manage credit based billing.
Cartesia features
- Sonic model
- Cartesia Sonic streams synthetic speech with a focus on low latency for live voice agents.
- Line platform
- Cartesia Line bills agent calls at $0.03 a minute, and a Cartesia number adds $0.014 more.
- Credit plans
- Cartesia Pro costs $5 for 100,000 credits and Scale $299 for 8 million, at about one credit per character.
- Overage switch
- Cartesia blocks speech when credits run out unless overage is turned on.
Against Google Speech to Text's pain points: Cartesia
Cartesia replaces Google voices in a real time agent, yet offers nothing to take over Google transcription.
- Transcription price: Cartesia sells speech output and agent minutes, so a Google transcription bill stays in place unless another engine replaces it
- Cloud setup: Cartesia plans start from a $5 Pro account, skipping the Google Cloud project and billing account setup
- Tough audio: Cartesia is a voice output product, so noisy call transcription is not something Cartesia claims to solve
- Voice cloning: Cartesia cloning terms are not stated on the pricing sources checked, so confirm the clone tier before choosing Cartesia for a branded voice
- One vendor: Cartesia splits spend between speech credits and agent minutes, so a monthly forecast has two moving parts
- Free credit: Cartesia Free includes 20,000 credits, roughly 20,000 characters of speech at one credit each
Cartesia pros and cons
Pros
- Reviewers single out low latency on Sonic, the main reason agent builders pick it over Google Neural2 or Chirp 3 HD voices.
- A $5 Pro plan with 100,000 credits makes testing cheap, and Scale at $299 for 8 million credits lowers the per character cost at volume.
Cons
- Billing splits between credits for speech and per minute agent fees, so forecasting a month takes more work than one Google line item.
- Speech stops when credits run out unless overage is switched on, a real risk for production traffic.
- Credit based billing with LLM tokens charged separately makes a talkative agent's cost harder to predict, and speed claims from the vendor should be checked on your own text.
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 run live agents that need streaming speech output and can manage credit based billing.
AssemblyAI
Built for accuracy focused transcription with audio intelligence add ons
AssemblyAI is a transcription API that developers shortlist for accuracy, with speaker labels, entity detection, sentiment and translation sold as per hour add ons on top of its Universal models.
Universal-2 batch costs $0.15 an hour, about $0.0025 a minute, and Universal-3.5 Pro $0.21 an hour. Speaker diarization adds $0.02 an hour and entity detection $0.08. New accounts get $50 of credit with no credit card, and there is no text to speech.
The add on model is the catch: speaker diarization is $0.02 an hour and entity detection $0.08, so an analytics heavy transcript costs more than the $0.15 Universal-2 base. With no voice output, AssemblyAI replaces Google Speech to Text but not its spoken replies. AssemblyAI suits developers who need transcripts plus analysis and can live without voice output.
AssemblyAI features
- Universal-2
- AssemblyAI prices Universal-2 batch audio at $0.15 per hour, near a quarter of a cent per minute.
- Universal-3.5 Pro
- AssemblyAI Universal-3.5 Pro costs $0.21 an hour for higher accuracy work.
- Speaker labels
- AssemblyAI diarization adds $0.02 an hour on top of the base model.
- Entity detection
- AssemblyAI entity detection adds $0.08 an hour, with sentiment and translation sold the same way.
- Signup credit
- AssemblyAI hands new signups $50 in credit, and no credit card is required.
- Review base
- AssemblyAI averages 4.6 over 125 G2 reviews, where accuracy and ease of use earn the most praise.
Against Google Speech to Text's pain points: AssemblyAI
AssemblyAI undercuts Google STT on rate, though add ons grow the hourly total.
- Transcription price: AssemblyAI charges about $0.0025 a minute on Universal-2, close to one sixth of what Google V2 bills at $0.016 for the first tier
- Cloud setup: AssemblyAI issues an API key on signup, which avoids the Google Cloud project, IAM and service account steps
- Tough audio: AssemblyAI is shortlisted for accuracy on G2, though reviewers also flag slow processing on some jobs, so test noisy files
- Voice cloning: AssemblyAI has no text to speech at all, so a branded voice needs another vendor
- One vendor: AssemblyAI covers transcription and analysis, so a team dropping both Google speech APIs still adds a voice provider
- Free credit: AssemblyAI $50 of credit covers about 333 hours of Universal-2 batch audio at $0.15 an hour
AssemblyAI pros and cons
Pros
- Universal-2 at about $0.0025 a minute is roughly a sixth of the Google V2 entry rate of $0.016.
- Diarization, sentiment and entity detection come from the same API, so call analysis needs no second service.
- G2 lists 125 reviews at 4.6, with accuracy and ease of use named most often.
Cons
- Each add on bills per hour on top of the base rate, so a full analysis stack lands well above the $0.15 headline.
- Speech to text only: a team replacing Google Text to Speech as well needs a second vendor for voice output.
- G2 reviewers flag limited language support and slow processing on some jobs, which matters for multilingual Google users.
AssemblyAI reviews
AssemblyAI holds 4.6 on G2 from 125 reviews. AssemblyAI on G2 ↗
A reviewer highlights: "It was quick for our team to integrate, and it delivers accurate transcription results even with long audio files and conversations involving multiple speakers." (Verified user, Mid-Market, G2 Apr 2026)
Another flags: "Transcription accuracy occasionally drops with poor audio quality or strong background noise." (Gaurav R., Small-Business, G2 Jun 2026)
On pricing: "Pricing is fair, and they give you a generous free trial to test your integration without having to pay." (Verified user in film, Small-Business, G2 Aug 2026)
Choose AssemblyAI if you need transcripts plus analysis and can live without voice output.
Deepgram
Built for streaming transcription plus Aura voices from one developer account
Deepgram replaces both Google speech APIs from one account, because one Deepgram account runs Nova-3 transcription, Aura-2 voices and a Voice Agent API. Billing is per second with no round up, and enterprise buyers can self host.
Nova-3 streaming lists $0.0077 a minute on pay as you go, and pre recorded audio $0.0043. Aura-2 costs $0.030 per 1,000 characters and the Voice Agent API starts at $0.075 a minute. Pay as you go accounts get $200 of credit that does not expire.
Running recognition, voices and the agent on one Deepgram account trades a higher agent minute for a single bill and one vendor contract. Deepgram fits developers who want one non Google vendor for transcription, voices and agents.
Deepgram features
- Nova-3
- Deepgram Nova-3 lists $0.0077 a minute streaming and $0.0043 pre recorded on pay as you go.
- Aura-2 voices
- Deepgram Aura-2 text to speech costs $0.030 per 1,000 characters.
- Voice Agent API
- Deepgram prices its Voice Agent API from $0.075 per minute.
- Per second billing
- Deepgram bills to the second with no round up, which keeps short clips cheap.
- Self hosting
- Deepgram offers self hosting to enterprise buyers.
Against Google Speech to Text's pain points: Deepgram
Deepgram mirrors both Google speech APIs in one account without a GCP project.
- Transcription price: Deepgram pre recorded audio at $0.0043 a minute is about 27% of the Google V2 entry rate, while streaming at $0.0077 is about half
- Cloud setup: Deepgram runs as a standalone account with an API key, so no Google Cloud project or IAM roles are needed
- Tough audio: Deepgram G2 reviewers name accuracy and real time transcription most often across 482 reviews at 4.6
- Voice cloning: Deepgram Aura-2 has no self serve voice cloning, so a branded voice still needs another route
- One vendor: Deepgram runs transcription, Aura-2 voices and agents under one key, replacing the two separate Google APIs
- Free credit: Deepgram $200 of non expiring credit covers about 46,500 minutes of pre recorded Nova-3 at $0.0043
Deepgram pros and cons
Pros
- One account covers transcription, text to speech and agents, mirroring the two Google APIs without a GCP project.
- $200 of non expiring credit buys about 26,000 minutes of Nova-3 streaming.
- G2 shows 482 reviews averaging 4.6, with accuracy, speed and real time transcription named most often.
- Per second billing with no rounding up keeps short clips cheap.
Cons
- Streaming at $0.0077 a minute costs nearly twice the pre recorded rate, and add ons such as diarization or redaction stack about $0.0013 to $0.0020 a minute each.
- No self serve voice cloning on Aura-2, and G2 reviewers mention limited language support and cost at scale.
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 want one non Google vendor for transcription, voices and agents.
Microsoft Azure Speech
Built for teams that want a second hyperscaler for speech instead of GCP
Microsoft Azure AI Speech is the main alternative cloud, covering speech to text and neural text to speech with many languages, for teams that want a large provider other than Google.
The free tier includes 5 hours of speech to text and half a million text to speech characters each month. The Azure pricing page showed no confirmable USD rate when checked on 21 October 2026, so price your volume in the Azure calculator before comparing.
The pull is organisational rather than price: a Microsoft shop gets speech on the cloud it already pays for, under one bill and existing identity management. Azure's missing public price is the weak spot, because volume has to be sized in Azure's own calculator before any honest comparison with Google. Azure Speech suits Microsoft teams that want a large provider other than Google.
Microsoft Azure Speech features
- Two services
- Azure AI Speech offers transcription and neural voices for many languages from one Microsoft service.
- Free tier
- Azure gives 5 hours of transcription and 0.5 million text to speech characters free each month.
- Custom Neural Voice
- Azure sells Custom Neural Voice as a paid, gated add on.
- Microsoft fit
- Azure suits teams already running Microsoft 365, per G2 reviewer comments on integrations.
Against Google Speech to Text's pain points: Microsoft Azure Speech
Azure swaps one large cloud for another, with similar setup overhead to Google.
- Transcription price: Azure showed no confirmable USD rate on its pricing page on 21 October 2026, so price volume in the Azure calculator before comparing with Google
- Cloud setup: Azure still needs subscriptions, resource keys and regions, a setup burden close to a Google Cloud project
- Tough audio: Azure G2 reviewers complain about accent recognition and noisy audio across 69 reviews at 3.9
- Voice cloning: Azure Custom Neural Voice is paid and gated, much like the Google Instant Custom Voice
- One vendor: Azure AI Speech bundles transcription and voices under one Azure resource, though both sit inside a full cloud account
- Free credit: Azure free tier gives 5 transcription hours a month, against 60 free Google minutes
Microsoft Azure Speech pros and cons
Pros
- A free tier of 5 transcription hours and 0.5 million text to speech characters a month gives more trial room than 60 free Google minutes.
- G2 reviewers praise accuracy, multilingual support and integrations, useful for teams already on Microsoft 365.
Cons
- G2 shows 3.9 from 69 reviews, with complaints about accent recognition, noisy audio and thin documentation.
- Azure is still a cloud account with subscriptions, resource keys and regions to manage, so the setup burden is close to GCP.
Choose Microsoft Azure Speech if you run a Microsoft shop and want speech on the cloud you already pay for.
Speechmatics
Built for accent heavy and multilingual transcription billed by the hour
Speechmatics is a UK transcription company known for accent and dialect coverage, with batch and real time models, a Flow voice agent product and newer text to speech. Speechmatics suits Google users whose speakers vary widely in accent.
Melia-1 costs $0.24 an hour in batch, about $0.004 a minute, and $0.40 an hour in real time. Standard runs $0.45 an hour and Enhanced $0.75 in batch. New accounts get $100 of free credit.
Real time transcription costs $0.40 an hour against $0.24 in batch, so live captioning runs dearer than offline jobs. A Flow voice agent and newer text to speech sit alongside the core recogniser. Speechmatics fits products whose speakers span many accents and dialects.
Speechmatics features
- Melia-1
- Speechmatics charges $0.24 per hour for Melia-1 batch jobs and $0.40 per hour for live audio.
- Accent coverage
- Speechmatics is known for accent and dialect coverage across its batch and real time models.
- Accuracy tiers
- Speechmatics batch pricing climbs to $0.45 per hour on Standard and $0.75 per hour on Enhanced.
- Flow agents
- Speechmatics Flow is a voice agent product next to its newer text to speech.
- Signup credit
- Speechmatics new accounts get $100 of free credit.
- Rate changes
- Speechmatics revised prices at least three times during 2026, so a quote from January needs rechecking.
Against Google Speech to Text's pain points: Speechmatics
Speechmatics targets the accented audio that Google STT reviewers struggle with.
- Transcription price: Speechmatics batch audio on Melia-1 works out near $0.004 a minute, one quarter of the $0.016 Google V2 charges first
- Cloud setup: Speechmatics runs as its own account with $100 of credit, outside any Google Cloud project
- Tough audio: Speechmatics focuses on accent and dialect coverage, and its 75 G2 reviews average 4.7 with accuracy named often
- Voice cloning: Speechmatics sells no self serve clone, so a custom brand voice needs a separate provider
- One vendor: Speechmatics now adds text to speech and Flow agents beside transcription, so one account can cover both directions
- Free credit: Speechmatics signup credit of $100 pays for roughly 416 batch hours on Melia-1
Speechmatics pros and cons
Pros
- Melia-1 batch at about $0.004 a minute is a quarter of the Google V2 entry rate.
- $100 of free credit covers about 416 hours of Melia-1 batch audio before any bill.
- G2 shows 4.7 from 75 reviews, against 4.6 for AssemblyAI and Deepgram, with accuracy and ease of use praised.
Cons
- Speechmatics changed its rates three or more times in 2026, so a quote from early in the year may not hold.
- Enhanced accuracy costs $0.75 an hour in batch, about three times the Melia-1 rate.
- G2 reviewers mention limited features and slow performance on some jobs.
- Speechmatics has no self serve voice cloning for a branded voice.
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 your products have speakers who span many accents and dialects.
Google Cloud Speech alternatives by job and use case
Match the pick to the part of Google Cloud Speech you are replacing.
Picks for each speech job
| Job | Pick | Why |
|---|---|---|
| Cheap transcription | PyAI Hear | Lists $0.001 a minute, a sixteenth of the Google V2 entry rate |
| Transcripts with analysis | AssemblyAI | Diarization, sentiment and entities as per hour add ons |
| Expressive voice output | ElevenLabs, PyAI Speak | Studio style narration, or free cloning at $0.04 a minute |
| Both STT and TTS | PyAI, Deepgram, Azure, ElevenLabs | One account covers speech in and out |
| Accented speakers | Speechmatics | Melia-1 at $0.24 an hour with wide accent coverage |
| Non GCP cloud | Azure Speech | 5 free STT hours a month on another hyperscaler |
Speech to text by price
| Tool | Price per minute |
|---|---|
| PyAI Hear | $0.001, metered per minute |
| AssemblyAI Universal-2 | About $0.0025 |
| Deepgram Nova-3 | $0.0043 batch, $0.0077 streaming |
| Speechmatics Melia-1 | About $0.004 |
| Google STT V2 | $0.016 up to 500,000 minutes |
Open source speech models
| Model | License | Note |
|---|---|---|
| Whisper | MIT | Transcribes locally, often cited for messy English batch audio |
| Chatterbox | MIT | Clones a voice from a short reference clip |
| GPT-SoVITS | Open | Few shot cloning from a seconds long sample |
| the flow-matching voice-cloning model | Open | Non autoregressive synthesis for quick generation |
| VoxCPM | Apache 2.0 | Multilingual synthesis under a permissive license |
Free and open source Google Cloud Speech alternatives
Google Cloud Speech has its own free tier: 60 minutes of transcription a month, plus 4 million Standard or WaveNet characters and 1 million Neural2 characters of text to speech, and new GCP accounts get $300 of credit that expires after 90 days. For self hosting, Whisper transcribes on your own hardware, while Chatterbox, GPT-SoVITS and the flow-matching voice-cloning model synthesize or clone voices for the price of a GPU. Among hosted APIs, signup credit makes a test simple: $50 on PyAI with no credit card, $50 on AssemblyAI, $100 on Speechmatics, $200 on Deepgram, 20,000 Cartesia credits and 10,000 ElevenLabs credits a month.
When Google Cloud Speech is still the right pick
Google Cloud Speech still makes sense for teams already running on Google Cloud that want transcription and voices on one invoice, broad language coverage, and Dynamic Batch at about $0.003 a minute for jobs that can wait up to 24 hours. G2 reviewers also credit it for ease of use and real time transcription. Teams that want a lower streaming rate, free cloning, or an API without a Google Cloud project gain the most from switching.
How to switch from Google Cloud Speech in 4 steps
Run the switch one workload at a time so a quality drop never reaches every user at once.
Step 1 of 4
Pick the API
Decide whether to replace Speech to Text, Text to Speech or both, and prefer an OpenAI compatible format so the code change stays small.
Step 2 of 4
Test your own audio
Run an hour of real recordings, including phone and accented audio, through the new API and through Google, then compare word error and turnaround.
Step 3 of 4
Compare cost at volume
Multiply each per minute rate by your monthly minutes, then add the diarization or other add ons you actually use.
Step 4 of 4
Switch one workload
Point one product or audio type at the new key first, keep Google as a fallback, and move the rest once accuracy holds.
Which Google Cloud Speech alternative to pick
Choose by the half of Google Cloud Speech you are replacing. For transcription, AssemblyAI adds analysis, Deepgram adds streaming and Speechmatics covers accents. For voice output, ElevenLabs suits expressive narration and Cartesia live agents. PyAI runs Hear at $0.001 a minute beside Speak with free cloning, under one key and without a Google Cloud project. A sandbox key costs nothing and carries $50 of credit.
More speech to text comparisons
Google Cloud Speech alternatives FAQs
Common questions from teams comparing Google Cloud Speech.
How much does Google Cloud Speech cost?
Speech to text from $0.016 a minute, text to speech from $4 per million characters. Speech to Text V2 is $0.016 a minute, falling to $0.004 at high volume. Text to Speech runs $4 per million characters on Standard and WaveNet, rising to $16 on Neural2, $30 on Chirp 3 HD and $160 on Studio.
What is a good Google Cloud Speech alternative?
AssemblyAI or Deepgram for transcription, ElevenLabs for voices. AssemblyAI pairs Universal-2 at about $0.0025 a minute with analysis add ons, Deepgram streams Nova-3 at $0.0077, and Speechmatics handles accents. For voice output, ElevenLabs suits narration and Cartesia live agents. PyAI covers both sides, with Hear at $0.001 a minute and free cloning on Speak.
What is cheaper than Google Cloud Speech to Text?
PyAI, AssemblyAI, Speechmatics and Deepgram. PyAI Hear lists $0.001 a minute, AssemblyAI Universal-2 about $0.0025, Speechmatics Melia-1 about $0.004 and Deepgram Nova-3 pre recorded $0.0043, all under the Google V2 entry rate of $0.016. Google reaches $0.004 only past 2 million minutes a month, or about $0.003 on Dynamic Batch for jobs that can wait.
Do I need a Google Cloud account to use these alternatives?
No. Google Cloud Speech runs inside a Google Cloud project with billing, IAM roles and service accounts. The alternatives here are standalone APIs with their own keys, so a team can skip the Google Cloud setup. PyAI keys work with an OpenAI compatible format, and new accounts open with $50 of credit and no credit card.
Does Google Cloud Speech do voice cloning?
Only a paid custom voice. Google offers Instant Custom Voice at $60 per million characters and a managed custom voice, not a free self serve clone. PyAI enrolls a cloned voice for free from a short clip, ElevenLabs offers instant cloning from its $6 Starter plan, and Chatterbox clones open source on your own GPU.
Which alternative covers both speech to text and text to speech?
PyAI and Deepgram. PyAI runs Hear for transcription and Speak for text to speech under one key, and Deepgram pairs Nova with Aura. Azure and ElevenLabs cover both too, while AssemblyAI and Cartesia each focus on one side, so mixing two vendors is common.
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.