# Speechmatics Features, Functionality and Deployments Source: https://www.speechmatics.com/product/features-and-deployments Speech-to-text API features: batch and real-time transcription, diarization, translation, and summarization. Deploy in the cloud, on-prem, or on-device. ## A speech-to-text API with three ways to deploy it: cloud, on-prem, or on-device. Batch and real-time transcription, speaker diarization, translation, and summarization, in one API. Built for voice agents, contact center tools, medical transcription and more. - [Speak to sales](https://www.speechmatics.com/speak-to-sales) ## Processing modes Get the very best performance whether you're transcribing pre-recorded files or live audio. Get the very best performance whether you're transcribing pre-recorded files or live audio. ### Configuration — Batch Transcription Transcribe pre-recorded files Process thousands of hours of pre-recorded files, whenever you need them, and fast. - [Learn more](https://docs.speechmatics.com/speech-to-text/batch/quickstart) ### Configuration — Real-Time Live audio transcription without compromise Transcribe media as it happens. Get **initial transcriptions in milliseconds**, with context-driven accuracy improvements over time. - [Learn more](https://docs.speechmatics.com/speech-to-text/realtime/quickstart) ## Deployment options Run Speechmatics wherever your infrastructure and compliance needs require Choose how Speechmatics runs — in the cloud, on your own infrastructure, or directly on-device — deployed however suits you. ### Configuration — On-Prem Deployment Meet architecture, [security](https://www.speechmatics.com/security) and compliance needs by hosting our API in [your own environment](https://www.speechmatics.com/enterprise). Deploy with [Docker containers, Kubernetes, or a preconfigured virtual appliance](https://docs.speechmatics.com/deployments). - [Learn more ](https://docs.speechmatics.com/deployments) ### Configuration — Cloud Deployment Get secure and scalable access to our API through our cloud deployment and get instant access to all our new features, languages and updates. - [Learn more](https://docs.speechmatics.com/deployments/) ### Configuration — On-Device Deployment Run Speechmatics directly on your devices for ultra-low latency and maximum data privacy — nothing sent over a network. Get cloud-parity accuracy (within 5%), full speaker diarization, and sub-second latency, on Mac or Windows. Ideal for use cases where connectivity is limited and data must stay local. - [On-Device](https://www.speechmatics.com/speech-to-text/on-device) ## Models Every model runs on the same API. Pick the one that matches your priorities. Melia 1: automatic language detection, labelling, and seamless code-switching in one API, across all 56+ supported languages. Available now for batch transcription, with real-time support on the way. #### Model overview | **Model** | **Best for** | **Real-time** | **Batch** | **Multilingual** | | --- | --- | --- | --- | --- | | Enhanced | Highest accuracy, one language per file or stream | ✓ Yes | ✓ Yes | ✗ No | | Standard | Strong accuracy with faster turnaround and lower cost | ✓ Yes | ✓ Yes | ✗ No | | Melia 1 | Audio that switches between languages mid-conversation, across all 56+ supported languages | Coming soon | ✓ Yes | ✓ Yes | #### Feature support by model | **Feature** | **Enhanced** | **Standard** | **Melia 1** | | --- | --- | --- | --- | | Speaker & channel diarization | ✓ Yes | ✓ Yes | ✓ Yes | | Word-level timestamps & punctuation | ✓ Yes | ✓ Yes | ✓ Yes | | Custom dictionary | ✓ Yes | ✓ Yes | Not yet | | Confidence scores | ✓ Yes | ✓ Yes | Not yet | | Translation, summaries, chapters | ✓ Yes | ✓ Yes | Not yet | | Code-switching across 56+ languages | ✗ No | ✗ No | ✓ Yes | **Diarization ships on every model, including Melia 1.** It's one of the areas Speechmatics is strongest, and it's not something every multilingual model gets right. See [pricing](https://www.speechmatics.com/pricing) for model costs, or the [developer docs](https://docs.speechmatics.com/speech-to-text/models) for configuration syntax. ## Transcription Features Everything you need to hit the highest accuracy possible Our customization options allow you to finely tune your set up to achieve [high accuracy](https://www.speechmatics.com/product/transcription) with even the most unique words and phrases. - [Transcription](https://www.speechmatics.com/product/transcription) ## Feature — Custom Dictionary [Boost accuracy](https://docs.speechmatics.com/speech-to-text/features/custom-dictionary) for proper nouns, acronyms or industry-specific terms by providing a list of custom words. ## Feature — Speaker & Channel Diarization Track who said what and when with [speaker labelling](https://docs.speechmatics.com/speech-to-text/features/diarization) for each word, available for both batch and real-time transcription. ## Feature — Numeral Formatting Identify and correctly format [numbers, dates and currencies](https://docs.speechmatics.com/speech-to-text/formatting#output) automatically to improve transcript readability and enable effective post-processing. ## Feature — Profanity & Disfluency Detection Aid comprehensibility and compliance by detecting and optionally removing words that are considered [profanities or hesitations](https://docs.speechmatics.com/speech-to-text/formatting#profanities). ## Features — File Formats Minimize the resource needed to prepare audio or video files with [support for all major audio and video formats](https://docs.speechmatics.com/speech-to-text/batch/input) along with automatic sample rate detection. ## Advanced Features Easily push a variety of media formats to the API Easily push a variety of media formats to the API and get a rich set of metadata to support your post processing needs. ## Features — Confidence Scores Collect [confidence scores](https://docs.speechmatics.com/speech-to-text/batch/output) for every word in the transcript to enable efficient human review and editing. ## Features — Word Timings Get [accurate timestamps](https://docs.speechmatics.com/speech-to-text/batch/output) for every word in the transcript to allow for post-processing and improved end user experience. ## Feature — Advanced Punctuation & Casing Improve readability with language-specific [capitalization and punctuation](https://docs.speechmatics.com/speech-to-text/formatting#punctuation) including commas, question marks and exclamation marks. ## Features — Audio Events Improve accessibility & fully-automate tedious captioning by identifying and labelling non-speech sounds in media, using AI. - [audio events](https://www.speechmatics.com/product/audio-events) ## Languages Partner with Speechmatics to maximize your total addressable market We deliver for multilingual, multicultural and multinational businesses, with coverage of 56+ languages across a range of dialects and accents. Accuracy tuning extends to specific industries too, including finance terminology and a dedicated [medical model reaching 93% accuracy](https://www.speechmatics.com/use-cases/medical-transcription) with 50% fewer keyword errors than the next-best system. - [Translation](https://www.speechmatics.com/languages) ## Language Coverage [56+ languages](https://www.speechmatics.com/languages), with code-switching at the word level. ## Accents and dialects Whether you need Brazilian Portuguese or Canadian French, we have you covered with a single language model that [supports all associated accents and dialects](https://docs.speechmatics.com/speech-to-text/languages). ## Translation Transcribe and [translate audio](https://www.speechmatics.com/product/translation) to and from English for over 30 languages using a single API call. ## Language Identification Simplify integration and ensure accurate transcription with [automatic detection](https://docs.speechmatics.com/speech-to-text/batch/language-identification) of the language spoken. ## Speech understanding [Transcription](https://www.speechmatics.com/product/transcription) pairs with [**summaries, sentiment, topics, and chapters**](https://docs.speechmatics.com/speech-to-text/batch/speech-intelligence/auto-chapters) in the same API call. ## Translation With [automatic translation](https://www.speechmatics.com/product/translation) with a single API call, you can translate media and provide captions for over half the world’s population. - [Translation](https://www.speechmatics.com/languages) ## Summaries Instantly generate summaries for social and video platforms, so viewers know what to expect, without you having to manually write. - [Summarization](https://www.speechmatics.com/product/summarization) ## Sentiment Don’t just rely on reviews. See how customers are feeling about every aspect of your service by identifying sentiment throughout calls. - [Sentiment](https://www.speechmatics.com/product/sentiment) ## Topics Your audience don’t want to (always) watch long media. Give them the topics discussed and the timestamps so they can engage with what they are most interested in. - [Topics](https://www.speechmatics.com/product/topics) ## Chapters As well as being divided up and summarized, each chapter is given a heading, making it super easy to find the most engaging content. - [Chapters](https://www.speechmatics.com/product/chapters) ## Certifications & compliance Built to meet enterprise security and compliance requirements Speechmatics meets the security and compliance bar enterprises need, wherever you deploy. See [speechmatics.com/security](https://www.speechmatics.com/security) for full detail. ## Certifications — ISO/IEC 27001:2022 Accredited to the ISO/IEC 27001:2022 information security standard. ## Certifications — SOC 2 Type II SOC 2 Type II-certified, independently audited on security controls over time, not just a point in time. ## Certifications — HIPAA Fully compliant with the Health Insurance Portability and Accountability Act, for healthcare and clinical use cases. ## Certifications — GDPR Compliant with GDPR and other regional privacy directives. ## Certifications — Encryption Data encrypted at rest (AES-256) and in transit (TLS 1.2+). ## Certifications — Data residency Processing environments in the US, EU, or Australia, to meet data sovereignty requirements. ## **Ready to Understand Every Voice?** Sign up to our free speech-to-text SaaS Portal and we’ll guide you through the integration of our API. - [Speak to sales](https://www.speechmatics.com/speak-to-sales) - [Try Free](https://portal.speechmatics.com/signup/)