Deepgram tells new builds to use Flux TTS, and Flux TTS costs half again as much per character as Aura-2. On Pay as you go, Flux TTS is $0.0450 per 1,000 characters and Aura-2 is $0.030, as of November 2026, while Deepgram calls Aura-2 its "previous-generation model" with voices across seven languages. Aura-1, the English-only first generation, is $0.0150, so the cheapest Deepgram voice is also the oldest and the newest voice carries the highest rate.
For a team leaving Aura, the alternatives worth hearing are Bland Speech, Cartesia, Rime, Murf, and ElevenLabs. Bland Speech heads the list on sound, since Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, and its WebSocket takes text from a language model and returns mu-law audio at 8 kHz for a phone line. The text to speech hub compares the wider field, and alternatives to other providers are on the alternatives family page.
Why do teams look for Deepgram Aura alternatives?
A team looks for a Deepgram Aura alternative when Deepgram's own upgrade path costs more than the model it replaces, because Flux TTS, the model Deepgram recommends, carries Deepgram's highest per-character rate.
Deepgram labels Flux TTS "Recommended for all new builds" and serves it on /v2/speak, while Aura-2 and Aura-1 stay on /v1/speak. A team on Aura that follows Deepgram's advice is already changing endpoints, which makes it a natural point to hear other voices before committing to the higher rate.
Deepgram also runs a promotion on Flux TTS: $1 in credits for every $1 spent, up to $500, through 31 December 2026. A promotion that ends in six weeks is a poor basis for a budget, so the standing rates matter more. Growth requires $4,000 or more a year in prepaid credits and bills overage at the Growth rate plus 10%, charged weekly.
What should I check in a Deepgram Aura alternative?
A buyer weighing Deepgram Aura alternatives should check each vendor's first-audio claim and how it was measured, what a character costs on the model each vendor recommends, and whether the voice needs to be cloned.
First-audio claims are the vendor's own and aren't comparable. Deepgram says Flux TTS responses start in as low as 80ms at 24kHz under production load. Cartesia says under 90ms, Murf says under 100ms, ElevenLabs says a median of about 100ms for Eleven v4 Turbo excluding application and network latency, and Rime gives Coda v2 30 ms at the 50th percentile measured at the server without network, 96 ms in a one-request benchmark, and sub-200ms end to end on its cloud API. Bland publishes 199ms to first audio, the highest figure on this list.
Which Deepgram Aura alternative sounds most human?
Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, at 1365, and Deepgram wasn't in those results.
- #1BlandBland Speech v31365
- 6thCartesiaSonic 3.51096
- 12thElevenLabsEleven v3Near the bottom938
Ranks are among 13 AI voices, and each rival shows the highest listed voice available on it. Real human recordings scored 1500, at the right edge of this scale. Bars start at 600, not zero. Source: Design Arena's published results.
Cartesia's Sonic 3.5 ranked 6th at 1096 and ElevenLabs' Eleven v3 ranked 12th at 938. Rime and Murf weren't in the results. Cartesia's Sonic 3.6 launched on 27 August 2026 and ElevenLabs' Eleven v4 on 28 September 2026, after the results, and neither was scored in them.
Design Arena's listeners hear two recordings of the same text, blind, and pick the one that sounds more like a real person. Each model had one female and one male American English voice, and prompts mostly ran up to about 40 words, so the results speak to short phone-style turns in American English.
What are the best Deepgram Aura alternatives?
The best Deepgram Aura alternatives are Bland Speech, Cartesia, Rime, Murf, and ElevenLabs, with Bland Speech first because it pairs the realism result of 4 August 2026 with phone-ready audio and a flat rate with no plan to manage.
What these rows mean
- Published price: each vendor's own list rate with the model or plan named. Rime bills text to speech by the minute and converts at about 1,000 characters a minute.
- Published first-audio claim: each vendor's own figure, measured its own way, so the column is a list of claims and not a race. Deepgram's own claim is as low as 80ms.
- Audio Realism Bench: Design Arena's blind listening test of 4 August 2026, with rank among 13 AI voices and an Elo score, which is a rating built from head-to-head wins and losses. A vendor's newer models were not in those results.
1. Bland Speech
Bland Speech is a text to speech model that Bland runs itself, sold as a studio and an API. One endpoint streams over HTTP at POST /v2/tts, and a WebSocket at /v2/tts/ws takes text as a language model writes it, with a new context interrupting the turn that is playing. Output is mu-law at 8 kHz or PCM from 8 to 48 kHz. An OpenAI-compatible endpoint takes a base URL, key, and model change from a team that already calls OpenAI's speech API, and Bland has a first-party Pipecat service and a LiveKit Agents plugin for Python. The text to speech API page lists the endpoints.
Bland Speech suits a team that wants the voice to carry the call and will accept a published 199ms to first audio, higher than Deepgram's as-low-as 80ms. An instant clone needs one clean sample of about 10 seconds, and a professional clone is fine-tuned on 30 minutes or more of audio after Bland's consent and rights review. Every clone is private to the customer's organization. Bland publishes a migration guide for Deepgram beside guides for ElevenLabs, Cartesia, PlayHT, Resemble AI, Google, and Polly.
Pros
- Bland Speech v3 ranked first of 13 AI voices in Audio Realism Bench results of 4 August 2026
- A WebSocket that takes text as a language model writes it, with barge-in in the protocol
- Pipecat and LiveKit integrations, and an OpenAI-compatible endpoint
- One standard per-character rate, with no plan tiers
Cons
- More than a hobby project needs, because Bland Speech is built for production calls
Pricing: $0.04 per 1,000 characters ($40 per million), pay as you go with no monthly plan, and 50,000 free characters once for a new account. Enterprise pricing is custom, set with Bland's sales team. The Bland Speech page has a studio.
2. Cartesia
Cartesia sells Sonic 3.6 in 44 languages and says first audio arrives in under 90ms. Cartesia suits a team that wants a speed-first claim and a plan ladder: Free is $0 with 20,000 credits, Pro is $5 with 100,000, Startup is $49 with 1.25 million, and Scale is $299 with 8 million, at one credit per character. Output includes mu-law and A-law at 8 kHz, and a Cartesia number for voice agents costs $0.014 a minute for telephony.
Pros
- Under 90ms first-audio claim, with mu-law and A-law at 8 kHz
- Instant clones from 10 seconds on Pro, and free professional clone training
- SOC 2 Type II certified
Cons
- Credits, rollover limits, and overage settings to manage each month
Pricing: Overages cost $65, $45, and $38 per million credits on Pro, Startup, and Scale, and overages are opt-in. Enterprise is custom.
3. Rime
Rime sells conversational voice models for enterprise voice agents, IVR, and telephony: Coda is the flagship text to speech model and Mist v3 is the low-latency model, and Rime puts Mist v3's time to first audio at 37 ms at the 50th percentile. Rime suits a regulated phone line, because Rime is SOC 2 Type 2 and HIPAA compliant and says enterprise customers can run Rime in a dedicated VPC or fully on-premises.
Pros
- G.711 mu-law output with 8 kHz sampling for telephony
- $50 of usage free every month on the Free plan
- HTTP, WebSocket, and gRPC streaming, with word-level timestamps on Coda
Cons
- Voice cloning is an enterprise order trained on 30 to 60 minutes of recordings
- Coda serves 10 cloud API languages, and each voice serves one
Pricing: Flex has no monthly fee, with Mist v3 at $0.030 a minute, Coda v2 at $0.050, and Prism speech to speech at $0.080. Scale is $100 a month at 20% off every rate, and Enterprise is custom.
4. Murf
Murf sells Falcon 2.1 for streaming agents and Gen2 for studio voiceover, and says Falcon 2.1 reaches first audio in under 100ms. Murf suits a team that wants the lowest per-character rate on this list, $0.01 per 1,000 characters for Falcon 2.1, with purchased credit that never expires. Murf publishes official LiveKit and Pipecat integrations, and says Falcon 2.1 can run inside a customer's own AWS account through AWS Marketplace.
Pros
- $0.01 per 1,000 characters for Falcon 2.1, with a $10 free API credit and no card
- Official LiveKit and Pipecat integrations
- 150+ voices in 35 languages
Cons
- Falcon 2.1 voices speak in a single style, Conversational
- Instant voice cloning is available on Enterprise only
Pricing: Falcon 2.1 is $0.01 per 1,000 characters and Gen2 is $0.03. Enterprise is custom.
5. ElevenLabs
ElevenLabs sells text to speech, speech to text, and voice agents, and launched Eleven v4 and v4 Turbo on 28 September 2026. ElevenLabs suits a team that wants a model family covering 90+ languages and self-serve cloning: an instant clone starts on Starter at $6 a month, and a professional clone starts on Creator at $22. Output includes ulaw_8000 and alaw_8000, and ElevenLabs publishes a guide for sending generated audio through Twilio.
Pros
- Instant cloning from 10 seconds of audio on Eleven v4
- Mu-law and A-law output at 8 kHz
- BAAs for HIPAA customers on Enterprise
Cons
- Eleven v4 bills $0.08 per 1,000 characters, twice Eleven v4 Turbo's $0.04
- The standard TTS WebSocket doesn't take Eleven v3 or v4
Pricing: API usage is $0.08 per 1,000 characters for Eleven v4 and $0.04 for Eleven v4 Turbo and Flash. Plans run Starter $6 a month, Creator $22, Pro $99, Scale $299, and Business $990, with Enterprise custom.
Which Deepgram Aura alternative should I pick?
Leaving Deepgram Aura comes down to which number a team trusts for its callers, a vendor's millisecond claim or a blind listening result. For a voice agent where the sound of the voice decides whether callers stay on the line, Bland Speech is the Deepgram Aura alternative to pick, because Bland Speech v3 scored 1365 in the blind tests Design Arena published on 4 August 2026, 269 points above Cartesia's Sonic 3.5 and 427 above Eleven v3, and Bland Speech returns mu-law audio for phone lines.
Price runs the other way. At 10 million characters a month and list rates, Murf's Falcon 2.1 costs $100, Deepgram's Aura-2 $300, Cartesia's Scale plan $375 with overage, Bland Speech $400, ElevenLabs' Eleven v4 Turbo $400, Deepgram's Flux TTS $450, and Rime's Coda v2 on Flex $500 at its conversion of about 1,000 characters a minute. A team that mostly needs speed and a low rate has good reasons to stay with Deepgram or move to Murf, and a team that mostly needs callers to believe the voice has a reason to pay the Bland Speech rate.
Bland Speech vs Deepgram Aura and the Deepgram Aura review go deeper on Deepgram, Cartesia vs Deepgram Aura sets the two speed-first vendors side by side, and Bland Speech vs Deepgram vs Rime compares phone call voices.
Frequently asked questions
What is the best Deepgram Aura alternative?
Bland Speech is the best Deepgram Aura alternative for voice agents on phone calls, because Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, and the API takes text from a language model over a WebSocket and returns mu-law audio at 8 kHz.
Is there a cheaper Deepgram Aura alternative?
Yes, Murf's Falcon 2.1 costs $0.01 per 1,000 characters, below Aura-1's $0.0150 and Aura-2's $0.030. Bland Speech's $0.04 sits above Aura-2 and below Flux TTS at $0.0450, ElevenLabs charges $0.04 to $0.08 per 1,000 characters by model, and Cartesia and Rime bill by credit and by minute.
Which Deepgram Aura alternative publishes the lowest first-audio claim?
Rime says Coda v2 reaches first audio in 30 ms at the 50th percentile, measured at the server without network, which is lower than Deepgram's as-low-as 80ms, Cartesia's under 90ms, and Murf's under 100ms. Bland publishes 199ms, and each company measures its own way.
Does Bland Speech work with Pipecat or LiveKit?
Yes, Bland Speech has a first-party Pipecat service that is maintained by the Pipecat team, with classes for the WebSocket and HTTP endpoints, and a LiveKit Agents plugin for Python. Deepgram, Murf, ElevenLabs, and Rime have Pipecat and LiveKit integrations or guides as well.
Can a team move from Deepgram to Bland Speech?
Yes, Bland publishes a migration guide for Deepgram. Deepgram voice IDs can't be reused, so a team picks a Bland voice or clones one with consent, and a realtime client opens one WebSocket per conversation and sends new text under one context per turn.
