What do users say?

Deepgram has the best developer experience and pretty easy to use interface. They keep launching new models which benchmark really good against my internal data and the latency is really low which is always desirable. Review collected by and hosted on G2.com.

Okay, at times I have seen that they do not support certain languages at the accuracy that we wanted for our production use cases, like Swedish and Norwegian languages. I would love if they could give a clarity on how soon they are going to upgrade their models, so I will be able to plan my roadmap accordingly. Review collected by and hosted on G2.com.

One of the best when it comes to supporting my native language(Bulgarian). The transcriptions are always smooth and I am hard-pressed to find any errors. The free credit you get when you start also is a great way to get acquainted with the platform and determine whether or not it is for you. Review collected by and hosted on G2.com.

The pricing can be discouraging, especially at scale. This is also the reason I am not more active: I just can't justify the cost when I'm not generating revenue for my projects. The onboarding can also be very confusing but there are sufficient tutorials to guide you. Review collected by and hosted on G2.com.

I use Deepgram to turn speech into text, which helps me quickly understand and read what people say in meetings, calls, and videos. It saves me time because I don't have to write notes or listen to recordings again. I like that it quickly and accurately turns speech into text and works well with different voices and accents. Its accuracy and ability to handle different voices and accents really helped when I was processing multi-speaker meetings and recorded calls from global events, even with background noise or speakers with strong accents. Setting up Deepgram was straightforward. The API is well-documented, authentication was simple, and I was able to start transcribing audio in just a few hours without any complex configuration. I also integrate Deepgram with tools like Slack and custom apps using their API. Our team switched to Deepgram from another transcription service because we needed accuracy. Review collected by and hosted on G2.com.

It works really well overall, but it can sometimes struggle when the audio is very noisy or when multiple people are speaking over each other. Review collected by and hosted on G2.com.

I appreciate how fast Deepgram is when using it for real-time production applications, as even small delays can make a big difference. Deepgram's speed makes the end-user experience very good. The accuracy is another aspect I find impressive since it really matters a lot. I've tried other solutions, but they didn't provide the perfect accuracy I get with Deepgram. When making calls, the voice agent sounds clear and natural, which improves the overall experience. The response time is much faster, and conversations feel smoother and more interactive. The voice quality is very important to me, and Deepgram makes sure there are no breakouts, avoiding any irritation. During live calls, it talks like a person with interactive and engaging responses. Review collected by and hosted on G2.com.

It would be great if we can increase capacity of it, because for less number it works fantastic but when we increase the number voice quality gets breaks. It would be better if we can make 8 to 10 number of calls at a time, if we increase the number, voice gets break 4 to 5. Review collected by and hosted on G2.com.

Deepgram provides very accurate and fast speech-to-text transcription, even for long audio recordings and real-time streams. The API is easy to integrate into applications, and the documentation is clear and developer-friendly. I especially like the low latency for real-time voice recognition and the ability to handle different accents and noisy environments. It has been reliable and scalable for production use. Review collected by and hosted on G2.com.

While Deepgram performs very well overall, the pricing can become expensive when processing large volumes of audio. Some advanced configuration options and features also require deeper documentation to fully understand. Occasionally, accuracy can drop slightly with heavy background noise or highly specialized vocabulary. Review collected by and hosted on G2.com.

I like the AI text-to-speech and the video support. The text-to-speech voices partially meet my quality bar. Review collected by and hosted on G2.com.

It often doesn’t work right the first time I use it, so I need to provide more context and constantly revise it. The speech-to-text is only sometimes accurate: it struggles a lot when more than one person is speaking, sometimes doesn’t recognize accents, and can be incorrect depending on speaking speed, so I have to listen to the audio to make corrections. Latency is sometimes delayed. Review collected by and hosted on G2.com.

The speed is the standout feature. I'm using Nova-2 with dual WebSocket streams (mic + tab audio simultaneously) for a real-time voice coaching Chrome extension, and interim results come back fast enough to display a genuinely live rolling transcript. Word-level confidence scores, smart_format, vad_events, and utterance_end_ms all work exactly as documented and saved me a lot of custom logic. The $200 starter credit is also a great touch - it gave me real runway to prototype and validate the product before committing to a paid plan. Review collected by and hosted on G2.com.

Nothing major so far. The documentation could occasionally go deeper on edge cases around dual-stream WebSocket setups and silence/KeepAlive behavior, which required some trial and error to get right. But overall these are minor friction points in an otherwise smooth experience. Review collected by and hosted on G2.com.

The speech-to-text accuracy and punctuation are awesome compared with GCP and Azure STT. Pricing is also better than other services. Review collected by and hosted on G2.com.

Sometimes Nova 2 performs better than Nova 3, and Nova 3 still doesn’t support keywords. Also, the multi-language detection isn’t very accurate when you compare results across multiple streams. In my DilMesh app, I create a separate stream for each language that might be spoken. At our events we generally use Turkish and English, so the app creates two streams—one for Turkish and one for English—and then selects the final results based on the confidence response. Multi stream method works terrific when you compared with built-in multi language detection. Review collected by and hosted on G2.com.

I really like the outstanding accuracy of Deepgram. In our law firm, where precision is critical, it consistently delivers highly accurate transcriptions even with varied accents, legal terminology, and occasionally poor audio quality. The speed of Deepgram is also impressive; what used to take hours of manual work is now done in minutes, which helps us process evidence faster, especially with tight court deadlines. The initial setup was relatively straightforward, thanks to the clear and well-structured API documentation that made integration manageable with our existing workflow. Review collected by and hosted on G2.com.

Occasionally, Deepgram tends to hallucinate — generating words or phrases that were not actually spoken in the recording. In a legal context, this is a concern because inaccurate transcriptions can be misleading. It requires us to manually review and verify transcripts before relying on them, which adds some extra time back into our workflow. Review collected by and hosted on G2.com.

The model's accuracy and the extremely low real-time latency are excellent. The integration was quick and the overall performance in Spanish surpasses most alternatives. Review collected by and hosted on G2.com.

That it is hosted only in the USA and not in the EU is a negative point. Also, the API sometimes turns out to be complex. Review collected by and hosted on G2.com.