Speechmatics Pricing

It's an opt-in programme that takes 33% off Speech to Text rates. By opting in, you grant permission for your audio and transcripts to be used to help improve our models; however, opting in does not guarantee that your data will be used for training. We may, at our discretion, store and use some, all, or none of the data you've made available. The discount applies regardless of whether your data is ultimately stored and used.

The programme is off by default, can be reversed at any time, and applies only to future usage from the point of opt-in or subsequent opt-out. It doesn't apply to Text to Speech.

Volume discounts are automatically applied on any billable usage above 500 hours for each type of Speech-To-Text in a given month. For example, if you use 800 hours of Real-time enhanced accuracy and 400 hours of Real-time standard accuracy, you will be billed as follows:

• 500 hours Real-time enhanced accuracy at base rate

• 300 hours Real-time enhanced accuracy at 20% discount

• 400 hours Real-time standard accuracy at base rate

Additional discounts are available starting from 24,000 hours usage per year. Speak to us to find out more!

We bill Pro tier customers on the 1st of each month for the previous months usage. Costs are billed to the second, based on the cost per hour.

Billing for Enterprise customers is on a custom basis.

Yes, absolutely you can! You get 3,000 minutes (50 hours) of Speech-to-Text and 1 million characters of Text-to-Speech free per month to try our award-winning technology.

When you reach your free limit, simply add your credit card details in your account settings to use more.

We offer three proprietary transcription models, available to all customers:

• Enhanced: when accuracy matters most, our Enhanced model delivers our highest accuracy across all our languages.

• Standard: when you need strong accuracy but file turnaround time or cost control are the priorities. Note that Standard does not provide turnaround time benefits when using Realtime speech-to-text.

• Melia 1: when your audio contains more than one language, our Melia 1 model transcribes multilingual speech in a single transcript, including speakers who switch language mid-conversation, with no need to select a language. It matches Standard on accuracy and is available for Batch transcription.

Different models suit different jobs, so you can always choose the right one for the task at hand.

Our AI model supports 56+ languages for transcription, with 69 pairs supported for AI translation.

Transcription

Arabic - Bashkir - Basque - Belarusian - Bengali - Bulgarian - Cantonese - Catalan - Croatian - Czech - Danish - Dutch - English - Esperanto - Estonian - Finnish - French - Galician - German - Greek - Hebrew - Hindi - Hungarian - Indonesian - Interlingua - Irish - Italian - Japanese - Korean - Latvian - Lithuanian - Malay - Maltese - Mandarin (Traditional - & - Simplified) - Marathi - Mongolian - Norwegian - Persian - Polish - Portuguese - Romanian - Russian - Slovak - Slovenian - Spanish - Swahili - Swedish - Tamil - Thai - Turkish - Ukrainian - Urdu - Uyghur - Vietnamese - Welsh

Translation

Bulgarian - Catalan - Croatian - Czech - Danish - Dutch - English - Estonian - Finnish - French - Galician - German - Greek - Hindi - Hungarian - Indonesian - Italian - Japanese - Korean - Latvian - Lithuanian - Malay - Mandarin (Traditional - & - Simplified) - Polish - Portuguese - Romanian - Russian - Slovak - Slovenian - Spanish - Swedish - Turkish - Ukrainian - Vietnamese - Bokmål > Nynorsk

Multilingual

Arabic - Spanish - Mandarin - Malay - Tamil

We're certified to SOC 2 Type II, ISO/IEC 27001:2022, GDPR and HIPAA. Full details and additional documents (such as our DPA and Sub-BAA) are available in the Trust Centre.

If you have not opted in to the model training discount programme, Realtime audio is never stored, and Batch data auto-deletes after 7 days or sooner via the API, and your data is never used to improve our models.

Please feel free to email us at hello@speechmatics.com - we're here to help!