If you’ve been shopping for a speech-to-text (STT) solution for your business, you’re not alone. In our recent State of Voice Technology 2022 report, 99% of respondents said they viewed voice-enabled experiences as a critical part of their company’s future enterprise strategy. But the sheer number of options for speech transcription might be overwhelming if you aren’t familiar with the space-from Big Tech to open source options, there’s a ton of choices, with different price points and different feature sets to choose from. Although this diversity is great, it can also make it confusing when you’re trying to compare different options and pick the right solution for you.
In this blog post, we’re going to break down the various STT APIs available today, telling you their various pros and cons, and providing a ranking that we think accurately represents the current STT landscape. Before we get to the ranking, we’re going to break down exactly what a speech-to-text API is, the core features you’d expect a STT API to have, and some key use cases for speech-to-text APIs.
What is a Speech-to-Text API?
At its core, a speech-to-text application programming interface (API) is simply the ability to call a service to transcribe audio into speech. The STT service will take the provided audio file, process it using either machine learning or a set of tools that combines machine learning with rule-based approaches, and then provide a transcript of what it thinks was said.
Key Features of Speech-to-Text APIs
In this section, we’ll survey some of the most common features that STT APIs offer. The key features that are offered by each API differ, and your use cases will dictate your priorities and needs in terms of which features to focus on.
-
Accurate transcription – The most important thing, regardless of what you’re using STT for, is accurate transcription. If you’re getting back transcripts that look like MadLibs, it’s unlikely you’re going to get much business value from them. The absolute baseline accuracy for readable transcriptions is 80%.
-
Batch or pre-recorded transcription capabilities – Batch transcription won’t be needed by everyone, but for many use cases, you’ll want a service that you can send batches of files to to be transcribed, rather than having to do it one-by-one on your end.
-
Real-time streaming – Again, not everyone will need real-time streaming. However, if you want to use STT to create, for example, truly conversational AI that can respond to customer inquiries in real time, you’ll need to use a STT API that returns its results as quickly as possible.
-
Multi-language support – If you’re planning to handle multiple languages or dialects, this should be a key concern. And even if you aren’t planning on multilingual support now, if there’s any chance that you would in the future, you’re best off starting with a service that offers many languages and is always expanding to more.
-
Automatic punctuation & capitalization – Depending on what you’re planning to do with your transcripts, you might not care if they’re formatted nicely. But if you’re planning on surfacing them publicly, having this included in what the STT API provides can save you time.
-
Profanity filtering or redaction – If you’re using STT as part of an effort for community moderation, you’re going to want a tool that can automatically detect profanity in its output and censor it or flag it for review.
-
Topic detection – If you’re looking to process large volumes of audio in order to better understand what’s being discussed, a STT API that offers topic detection could be something you want to focus on.
-
Custom vocabulary – Being able to define custom vocabulary is helpful if your audio has lots of custom terms, abbreviations, and acronyms that an off-the-shelf model wouldn’t have been exposed to.
-
Keyword boosting – Similar to defining custom vocabulary, keyword boosting lets you make it more likely that the STT API will predict words that are particularly important or common in your audio.
-
Tailored models – If keyword boosting and custom vocabulary aren’t enough for your needs and you’re still seeing poor accuracy, you might want to look for a provider that will let you tailor a model for your specific needs, based on your own data. This typically improves accuracy beyond what any out-of-the-box solution can.
-
Accepts multiple audio formats – Another concern that won’t be present for everyone is whether or not the STT API can process audio in different formats. If you have audio coming from multiple sources that aren’t encoded in the same format, having a STT API that removes the need for converting to different types of audio can save you time and money.
Speech-to-Text Use Cases
As noted at the outset, voice technology that’s built on the back of STT APIs is a critical part of the future of business. So what are some of the most common use cases for speech-to-text APIs? Let’s take a look.
-
Smart assistants – Smart assistants like Siri and Alexa are perhaps the most frequently encountered use case for speech-to-text, taking spoken commands, converting them to text, and then acting on them.
-
Conversational AI – Voicebots let humans speak and, in real time, get answers from an AI. Converting speech to text is the first step in this process, and it has to happen quickly for the interaction to truly feel like a conversation.
-
Sales and support enablement – Sales and support digital assistants that provide tips, hints, and solutions to agents by transcribing, analyzing and pulling up information in real time. It can also be used to gauge sales pitches or sales calls with a customer.
-
Contact centers – Contact centers can use STT to create transcripts of their calls, providing more ways to evaluate their agents, understand what customers are asking about, and provide insight into different aspects of their business that are typically hard to assess.
-
Speech analytics – Broadly speaking, speech analytics is any attempt to process spoken audio to extract insights. This might be done in a call center, as above, but it could also be done in other environments, like meetings or even speeches and talks.
-
Accessibility – Providing transcriptions of spoken speech can be a huge win for accessibility, whether it’s providing captions for classroom lectures or creating badges that transcribe speech on the fly.
Top 9 Speech Recognition APIs
With that background out of the way, let’s dive into our ranking, and what we think are the top 9 speech-to-text APIs available today.
1. Deepgram Speech-to-Text API
-
Summary: We might be biased, but we think Deepgram is the best STT API on the market. We’re a developer-focused ASR provider with SDKs, providing an API that data scientists and developers can use to convert messy, unstructured audio data into accurate and structured transcriptions in batch or real-time-both on premises and in the cloud. Deepgram out of the box accuracies are in the 90%+ range with an option to customize speech models and reach even higher accuracies. Deepgram also has the fastest ASR in the market, with a 1200x real-time speed for batch processing and has less than a 300 millisecond lag on real-time streaming. If you’d like to give Deepgram a try, you can sign up for a free API key or contact us if you have questions.
-
Architecture: Built on the latest end-to-end deep learning neural networks
-
Pros:
-
Highest out-of-the-box and tailored model accuracy
-
Fastest speed
-
High customization within days
-
Easy to start with Console
-
-
Cons:
-
Fewer languages than big tech ASR, but we’re regularly releasing new languages
-
-
Price: $0.25/audio hour
