I really appreciate Deepgram for its speed, which is much faster than my previous transcription software and has been very reliable without any outages or API issues. The initial $100 worth of credits was great, allowing me to try it risk-free and see that it was good. The speed greatly improves its performance in production, and the API is really easy to use, making the initial setup very easy. Review collected by and hosted on G2.com.

I think the accuracy could be a little better, specifically the word accuracy of the transcription. Review collected by and hosted on G2.com.

Good transcript of audio. Affordable. Nice UI. Reliable API Review collected by and hosted on G2.com.

Didn’t notice for my tasks, frankly. Was not clear the difference between the models Review collected by and hosted on G2.com.

I use Deepgram for transcription in voice AI agents, and I really appreciate its low latency. The new flux transcriber is great for handling end of speech. It's state of the art, probably the best transcriber almost existing right now. The flux transcriber is well trained on conversation data, which makes it quite good in scenarios where there are longer pauses, like when something unexpected occurs. The latency of Deepgram was a big improvement for us, and the initial setup was easy. Review collected by and hosted on G2.com.

We had some outages in the past, which were quite terrible because then, like, the whole product doesn't work. Review collected by and hosted on G2.com.

I use Deepgram for Text to Speech and Speech to Text, and it solved the pronunciation problems for me. I like Deepgram's real-time streaming feature and the accuracy it provides. I also appreciate the high number of supported languages, which is extremely useful for developing a voice agent in vernacular languages of countries like India. The initial setup was easy and smooth, and I find it to be one of the best solutions. Additionally, switching from OpenAI Whisper to Deepgram was beneficial due to the real-time streaming, language support, and no additional infra overhead. Review collected by and hosted on G2.com.

understanding the Indian accent of the English Language. Review collected by and hosted on G2.com.

faster tranacription streaming. Better AI models quLITY Review collected by and hosted on G2.com.

a bit more expensive and poorer tts endpoint. voices could be ebtter Review collected by and hosted on G2.com.

I have used this awesome AI tool for speech to text and text to speech conversion. It helps to live translate things and make event and logs. I really like how fast it is in real-time, and its good noise cancellation features. It's also easy to integrate, and I enjoyed the real-time conversation features. What stood out to me the most is the flexibility, accuracy, and smart, fast performance compared to Google Speech to Text. The initial setup was pretty good and simple. Review collected by and hosted on G2.com.

Sometimes accent is the problem, but it is a very minor issue. Nothing specific otherwise, just more training on different accents. Review collected by and hosted on G2.com.

[AI] I like that Deepgram has all of the features I need, like multiple languages, low latency, and no training on our data, with a simple API. Deepgram is deeply integrated into my call center product and provides essential transcriptions. It's the first streaming transcription tech we tried that actually worked. The initial setup was very easy with the WebSocket API being pretty simple, and the authentication is simple too. Review collected by and hosted on G2.com.

Deepgram sometimes makes transcription errors, but this is understandable as no AI is perfect (especially with low latency). Deepgram can get pricy if you accidentally leave it on (e.g. we accidentally transcribed for 8 hours after a call ended; totally my mistake!). Review collected by and hosted on G2.com.

I mainly use Deepgram to quickly turn audio from calls and recordings into text without doing it manually. It saves me a lot of time by automatically transcribing audio, which reduces manual work and fits smoothly into my workflow without much setup. I like how fast and accurate it is and how easy it is to start using right away. It was too easy for me and my team to set up, and I switched to it because it's more effective than other tools. I would give it a 9 out of 10. Review collected by and hosted on G2.com.

Sometime accuracy drops with noisy audio or strong accents so improving that would make it even better. Review collected by and hosted on G2.com.

I appreciate Deepgram's speed and how it gives us decent results quickly. It's nice that I can configure keywords with the model, which is quite useful. I think it's pretty easy to work with, especially when the workload isn't too heavy. We found it to be generally the best and fastest for our Text to Speech needs, which are the features we care about most. The developer docs made it easy to work with, adding to its appeal. Using Deepgram for our voice interview platform is valuable because any latency reduction we can achieve is appreciated. Review collected by and hosted on G2.com.

Maybe just speed on the models and latency. Review collected by and hosted on G2.com.

I have used Deepgram for STT services and found it to be good. It helps me manage different pronunciations and accents, like those from US and India. The pricing model is much cheaper and the configuration is simple. I know how to configure and manage the streaming and more aspects of it. I also appreciate the pricing for usage, the rate limit, and the free credit for development processes. The documentation is proper, allowing even a fresher developer to set it up easily. Review collected by and hosted on G2.com.

Improve in TTS service, the latency in TTS is a little bit high. I have not used it that much for TTS, but as compared to Catesia, it was a little bit slow. Review collected by and hosted on G2.com.