# Deepgram Aura vs Speechma for scripts and voice-overs > Source: https://dubsheet.com/blog/deepgram-aura-vs-speechma-for-scripts-and-voice-overs/ Comparison [Text to speech and AI voice-over](https://dubsheet.com/category/text-to-speech) Pick Speechma for free browser voice-over; pick Deepgram Aura for a metered API that bills per character and adds team seats. [Dubsheet Research](https://dubsheet.com/authors/editorial) Published 3 Nov 2026 12 min read [Editorial policy](https://dubsheet.com/editorial-policy) ![Deepgram Aura vs Speechma.](https://dubsheet.com/img/1024/heroes/deepgram-aura-vs-speechma-for-scripts-and-voice-overs/b044e4441a2b/list-16x9-800.webp) On this page 1. [The short answer](#the-short-answer) 2. [Deepgram Aura and Speechma side by side](#deepgram-aura-and-speechma-side-by-side) 3. [What you pay and what the price counts](#what-you-pay-and-what-the-price-counts) 4. [What the free plan lets you publish](#what-the-free-plan-lets-you-publish) 5. [What happens when you run past your allowance](#what-happens-when-you-run-past-your-allowance) 6. [How much control you get over delivery and pronunciation](#how-much-control-you-get-over-delivery-and-pronunciation) 7. [Sharing the work and keeping your audio out of training](#sharing-the-work-and-keeping-your-audio-out-of-training) 8. [Each one in detail](#each-one-in-detail) 9. [How they compare](#how-they-compare) 10. [Which one to pick](#which-one-to-pick) 11. [How we compared](#how-we-compared) 12. [Questions buyers ask](#questions-buyers-ask) 13. [Sources](#sources) - [This page as Markdown](https://dubsheet.com/blog/deepgram-aura-vs-speechma-for-scripts-and-voice-overs.md) ## The short answer [Speechma](https://dubsheet.com/speechma) suits a creator who wants to publish voice-overs for free. The free audio may be monetized with no credit required. [Deepgram Aura](https://dubsheet.com/deepgram-aura) suits a developer or team putting speech into an app or a call, at $0.015 per 1k characters drawn from prepaid credit, with owner, admin and member roles at no seat fee. On 3 November 2026 we listed 53 tools: 42 of them under Free text to speech and 38 with a free plan. - Deepgram Aura charges $0.015 per 1k characters, metered on what you use. - Speechma has a free plan. - Deepgram Aura keeps running past the prepaid credits on its annual Growth plan, billed at the Growth rate plus 10%, while the Speechma API refuses requests once the monthly quota is used. - Deepgram Aura is the 1 of the 2 with team seats, and seats carry no separate fee. - Deepgram Aura is the 1 of the 2 that may use your content to train its models, though each API request can opt out. In this article - [Speechma](https://dubsheet.com/speechma) Free, ad-supported text to speech, cloning and dubbing in the browser with no account and commercial use included. - [Deepgram Aura](https://dubsheet.com/deepgram-aura) Per-character text to speech API from a speech-recognition company, tuned for real-time voice agents. ## Deepgram Aura and Speechma side by side Table 1 Price and the points where they differ, checked 2 November 2026 Deepgram Aura Speechma From Free, then $0.03 per 1k characters Free, then $1.50 a month Emotion, pace and emphasis Partly Yes Free plan or trial Yes Yes Works inside Yes Partly Voices that use more Yes No Past your allowance Yes No Pronunciation fixes Yes Partly Team seats Yes No Your voice trains their models Yes No Best for developers streaming speech into phone calls creators who need free voice-over [How we check](https://dubsheet.com/methodology/) ## What you pay and what the price counts Speechma wins on price, since every tool is free in the browser and paid plans start at $1.50 a month. SpeechmaPRO caps characters per conversion at 10,000, and the API meters characters per month, 1M on Starter and 50M on Pro. Deepgram Aura charges $0.015 per 1k characters of input text, metered on usage. The cost comes out of a prepaid dollar credit balance, not a monthly character allowance, so every character you send is billed. ## What the free plan lets you publish Speechma wins: every tool is free, with a 5-digit security code per generation and ads on the page. The free audio carries no watermark or required credit, and commercial use is included when you own the rights to what you put in. Deepgram Aura offers free use only in part. No paid plan gates commercial use: any customer may build the service into its own product, and Deepgram pitches marketing and advertising voice-overs, but reselling the service on its own is barred. ## What happens when you run past your allowance Deepgram Aura wins if your work must not stop. Pay as you go draws down prepaid credit at the listed rate with an optional auto-reload, and on the annual Growth plan usage past the prepaid credits is billed at the Growth rate plus 10%, charged weekly. Some Deepgram Aura voices use more credit than others. The Speechma API sells no extra characters past the monthly quota: requests are refused with a quota error until you move up to a larger plan. Every API plan includes the full voice catalog at one character rate, and premium voices use no more allowance. ## How much control you get over delivery and pronunciation Speechma wins on delivery and Deepgram Aura on pronunciation. Speechma’s Voice Effects set pitch, speed, volume, pauses and speaking styles, but its only pronunciation fix is respelling a word phonetically in the script each time. Deepgram’s Aura-2 model sets speaking speed but supports no inserted pauses and has no emphasis or style tags. It does accept inline pronunciation overrides for words, in English and Spanish. ## Sharing the work and keeping your audio out of training Deepgram Aura wins on sharing: the project owner invites team members as owner, admin or member, and seats carry no separate fee. Deepgram may use customer content to improve its models unless each API request opts out through a documented parameter. Speechma wins on privacy: your recordings are processed only to make the result, are not kept afterward, and partners may not use them for their own purposes. ## Each one in detail ### [Deepgram Aura](https://dubsheet.com/deepgram-aura) **Best for developers streaming speech into phone calls:** Deepgram Aura sends speech into calls through a documented Twilio guide and direct LiveKit support, billed by usage at $0.015 per 1k characters. - **Price:** Free, then $0.03 per 1k characters - **Free plan:** yes - **Who it’s for:** Developers building voice agents or customer service lines who want speech through an API at $0.015 per 1k characters, with no separate fee for teammates. - **Skip it if:** Creators who want to paste a script into a page and download audio should pick something else, because Aura has no project view and caps each request at 2,000 characters, so your own code has to split and rejoin a long script. Deepgram Aura is a text to speech API from a speech-recognition company, built for real-time voice agents. Official software development kits (SDKs) cover JavaScript, Python, Go and Java, and Deepgram documents direct use with LiveKit and a Twilio guide for streaming speech into calls. Downloads include MP3 and Opus, and live streaming offers only the raw telephony formats. Aura-2 lets you set speaking speed and override how words are pronounced in English and Spanish, but it does not support inserted pauses or emphasis tags. Aura costs $0.015 per 1k characters, metered on usage, and that is the regular pay-as-you-go rate, not a promotion. Pay as you go draws down prepaid credit with an optional auto-reload. On the annual Growth plan, usage past the prepaid credits is billed at the Growth rate plus 10%, charged weekly, where Speechma’s API refuses requests once the quota runs out. **What it does well** - Pay as you go at $0.015 per 1k characters, with a shared team project where each member gets an owner, admin or member role. - A Twilio guide and direct LiveKit support for streaming speech into calls. - Inline pronunciation overrides for words in English and Spanish. - Downloads as MP3 or Opus, plus the mu-law and A-law telephony formats phone systems use. **Where it falls short** - Each request is capped at 2,000 characters, so long scripts need your own code to split and rejoin them. - Deepgram may use customer content to improve its models unless each API request opts out through a documented parameter. - Aura-2 has no inserted pauses and no emphasis or style tags. ### [Speechma](https://dubsheet.com/speechma) **Best for creators who need free voice-over:** Speechma converts scripts free in the browser, and commercial use is included. - **Price:** Free, then $1.50 a month - **Free plan:** yes - **Who it’s for:** Creators and marketers who want free voice-over or dubbed video in the browser, with commercial use included. - **Skip it if:** Developers who may need more API characters than their monthly quota should pick something else, because Speechma’s API refuses requests past the quota and sells no extra characters. Its dubbing tool returns a downloadable MP4 with the speech replaced in another language, but the speaker’s lip movements stay unchanged. Every tool is free in the browser, with ads on the page and a 5-digit security code for each generation. Free dubbing accepts MP4 video files up to 60 MB in nine target languages. Longer scripts are split into separate clips and downloaded together as a ZIP, not managed as one project. Paid plans start at $1.50 a month, and SpeechmaPRO raises the cap to 10,000 characters per conversion. Premium voices use no extra allowance, and every API plan includes the full voice catalog at one character rate, while some Deepgram Aura voices use more. Speechma does not keep your recordings after processing, while Deepgram may use customer content to improve its models. **What it does well** - Every tool is free in the browser, with audio you may monetize. - Commercial use is included on the free tools. - Paid plans from $1.50 a month, and every API plan includes the full voice catalog at one character rate. - Your recordings are used to make the result you asked for, then discarded. **Where it falls short** - The API refuses requests past the monthly quota and sells no extra characters. - No pronunciation dictionary: a mispronounced word has to be respelled in the script every time. ## How they compare **Figure 1. Of 8 questions both answer, Deepgram Aura says yes to 6 and Speechma says yes to 2** | Product | Questions answered yes | | --- | --- | | Deepgram Aura | 6 of 8 | | Speechma | 2 of 8 | Source: Our answers to the same questions for both products, each read on the vendor’s own pages. Deepgram Aura’s 6 of 8 include voices that use more credit and content that may train its models, so a higher count is not a better score for someone reading scripts. Speechma’s 2 of 8 are yes on emotion, pace and emphasis and a free plan. > “Open the page and start creating. Nothing to install or pay.” From [Speechma’s own page](https://speechma.com/), read on 2 November 2026 ## Which one to pick - **If you need to stream generated speech into a LiveKit or Twilio call:** pick [Deepgram Aura](https://dubsheet.com/deepgram-aura). Deepgram Aura has an API and documents direct use with LiveKit and a Twilio guide for streaming generated speech into calls. - **If you need several people working in one shared project:** pick [Deepgram Aura](https://dubsheet.com/deepgram-aura). Deepgram Aura lets a project owner invite team members as owner, admin or member at no seat fee. - **If you need to dub a video or clone a voice as well as read a script:** pick [Speechma](https://dubsheet.com/speechma). Speechma dubs a video from 12 source languages into 9 target languages and returns an MP4, and clones a voice from a 5 to 30 second sample. - **If you need to set emotion, pace and emphasis:** pick [Speechma](https://dubsheet.com/speechma). Speechma’s Voice Effects set pitch, speed, volume, pauses and speaking styles, while Deepgram’s Aura-2 model sets speed but supports no inserted pauses and no emphasis or style tags. - **If you need your recordings kept out of model training:** pick [Speechma](https://dubsheet.com/speechma). Speechma does not keep your recordings after processing and its partners may not use them for their own purposes, while Deepgram may use customer content unless each API request opts out. ## How we compared We compared 2 tools on price, free plan and 39 questions about what they do, reading each vendor’s pricing page and product pages between 2 September 2026 and 2 November 2026. We did not sign up for or run any of them. The figures are as of 2 November 2026, from 2 listings. See [how we price an hour](https://dubsheet.com/methodology/) and [the price log for voice tools](https://dubsheet.com/price-changes). ## Questions buyers ask ### Can I use Speechma audio commercially for free? Yes. [Speechma](https://dubsheet.com/speechma) includes commercial use for generated audio, and the free audio carries no watermark or required credit and may be monetized. The condition is that you own the rights to the source text, recordings and videos you put in. ### How much does Deepgram Aura cost? [Deepgram Aura](https://dubsheet.com/deepgram-aura) costs $0.015 per 1k characters of input text, metered on usage. Charges come out of a prepaid dollar credit balance rather than a monthly character allowance. Team seats add no separate fee. ### Does Deepgram Aura have a free plan? Only in part. [Deepgram Aura](https://dubsheet.com/deepgram-aura) offers free use only partly, and [Speechma](https://dubsheet.com/speechma) is the 1 of the 2 with a free plan. Paid use of Deepgram Aura draws from prepaid credit at $0.015 per 1k characters. ### What happens when I hit the Speechma API limit? Your requests are refused with a quota error. The [Speechma](https://dubsheet.com/speechma) API sells no extra characters past the monthly quota, and the documented fix is moving up to a larger plan. ### Can my team share a Deepgram Aura or Speechma account? [Deepgram Aura](https://dubsheet.com/deepgram-aura), yes: the project owner invites team members and gives each an owner, admin or member role, with no per-seat fee. ### Does Deepgram use my content to train its models? It may. Deepgram may use customer content to improve its models, but each API request can be excluded through a documented parameter. [Speechma](https://dubsheet.com/speechma) does not keep your recordings after processing. ### Can I fix how a word is pronounced? In [Deepgram Aura](https://dubsheet.com/deepgram-aura), yes: the Aura-2 model accepts inline pronunciation overrides for words in English and Spanish. In [Speechma](https://dubsheet.com/speechma), only by respelling the word phonetically in the script each time, since there is no pronunciation dictionary for a whole project. ## Sources - [Deepgram Aura pricing page](https://deepgram.com/pricing), read on 1 November 2026 deepgram.com - [Deepgram Aura, deepgram.com/ai-voice-generator](https://deepgram.com/ai-voice-generator), read on 1 November 2026 - [Deepgram Aura, deepgram.com/product/text-to-speech](https://deepgram.com/product/text-to-speech), read on 1 November 2026 - [Deepgram Aura, deepgram.com/terms](https://deepgram.com/terms), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/docs/livekit-integration](https://developers.deepgram.com/docs/livekit-integration), read on 31 October 2026 - [Deepgram Aura, developers.deepgram.com/docs/text-to-speech](https://developers.deepgram.com/docs/text-to-speech), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com](https://developers.deepgram.com/docs/the-deepgram-model-improvement-partnership-program), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/docs/tts-media-output-settings](https://developers.deepgram.com/docs/tts-media-output-settings), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/docs/tts-models](https://developers.deepgram.com/docs/tts-models), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/docs/tts-voice-controls](https://developers.deepgram.com/docs/tts-voice-controls), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/docs/voice-agent-architecture](https://developers.deepgram.com/docs/voice-agent-architecture), read on 2 September 2026 - [Deepgram Aura, developers.deepgram.com/guides/deep-dives/managing-projects](https://developers.deepgram.com/guides/deep-dives/managing-projects), read on 1 November 2026 - [Deepgram Aura, developers.deepgram.com/support](https://developers.deepgram.com/support), read on 1 November 2026 - [Speechma pricing page](https://speechmapro.com/), read on 2 November 2026 speechmapro.com - [Speechma, speechma.com](https://speechma.com/), read on 2 November 2026 - [Speechma, speechma.com/ai-dubbing](https://speechma.com/ai-dubbing), read on 30 October 2026 - [Speechma, speechma.com/ai-voice-ethics](https://speechma.com/ai-voice-ethics), read on 2 November 2026 - [Speechma, speechma.com/api](https://speechma.com/api/), read on 2 November 2026 - [Speechma, speechma.com/contact](https://speechma.com/contact), read on 2 November 2026 - [Speechma, speechma.com/faq](https://speechma.com/faq), read on 2 November 2026 - [Speechma, speechma.com/guides/dubbing](https://speechma.com/guides/dubbing), read on 2 November 2026 - [Speechma, speechma.com/how-to-add-voiceover-in-canva](https://speechma.com/how-to-add-voiceover-in-canva), read on 26 October 2026 - [Speechma, speechma.com/how-to-make-an-audiobook-with-ai](https://speechma.com/how-to-make-an-audiobook-with-ai), read on 26 October 2026 - [Speechma, speechma.com/privacy](https://speechma.com/privacy), read on 2 November 2026 - [Speechma, speechma.com/terms](https://speechma.com/terms), read on 2 November 2026 About the author [Dubsheet Research](https://dubsheet.com/authors/editorial) Research desk The Dubsheet research desk publishes source-linked product research and original market analysis. Last checked 5 November 2026 ## Top of text to speech and AI voice-over [The category page](https://dubsheet.com/category/text-to-speech) - [Acapela Group](https://dubsheet.com/acapela-group) · Text to speech voices in over 30 languages, with SDKs, a cloud API and custom voice creation for apps and devices. · On quote Checked 29 Oct · [Visit](https://dubsheet.com/go/acapela-group) - [DupDub](https://dubsheet.com/dupdub) · AI voice-over, video dubbing, avatars and subtitles in one credit-metered studio. · From $15 a month Checked 5 Nov · [Visit](https://dubsheet.com/go/dupdub) - [ElevenLabs](https://dubsheet.com/elevenlabs) · Voice studio where one credit pool covers speech, cloning, dubbing, captions and audiobooks. · Free, then $6 a month Checked 1 Nov · [Visit](https://dubsheet.com/go/elevenlabs) - [Fish Audio](https://dubsheet.com/fish-audio) · Expressive text to speech directed with bracketed emotion tags, plus a community library of over two million voices. · Free, then $15 a month Checked 2 Nov · [Visit](https://dubsheet.com/go/fish-audio) - [Fliki](https://dubsheet.com/fliki) · Script-to-video studio with voice cloning, multilingual dubbing, and direct social publishing. · Free, then about $27 a seat a month Checked 1 Nov · [Visit](https://dubsheet.com/go/fliki) - [Luvvoice](https://dubsheet.com/luvvoice) · Browser text-to-speech with 200+ voices in 70+ languages, voice cloning, MP3 download and an API. · Free, then $8 a month Checked 28 Oct · [Visit](https://dubsheet.com/go/luvvoice) - All 27 text to speech and AI voice-over tools · First paid plans from $1.50 to $94 a month. · [See all 27](https://dubsheet.com/category/text-to-speech) - 20 free to start - 23 publish a monthly price - Checked 27 October–5 November 2026 [Show the whole category](https://dubsheet.com/category/text-to-speech) The Dubsheet digest One finding from our voice price data, every Tuesday. [Privacy policy](https://dubsheet.com/privacy) ## More of our research Buyer's guide 29 October 2026 ### [API text to speech: which tools sell it, on which plan](https://dubsheet.com/blog/api-text-to-speech-which-tools-sell-it-on-which-plan/) Typecast, Speechma, Fish Audio and Murf AI sell a speech API you can sign up for online, while Voicemaker and Narakeet offer one with conditions. Buyer's guide 27 October 2026 ### [Choosing an AI voice generator: meters, rights, cloning](https://dubsheet.com/blog/choosing-an-ai-voice-generator-meters-rights-cloning/) Pick an AI voice by the unit its price counts, what the free plan leaves on your audio, commercial rights and whether cloning needs consent.