What do you like best about the product?

As far as we tested it, Deepgram showed the most accurate timestamps. For our specific purposes, we need to timestamp all words elicited by feature films. In these movies, the speech, the music and the audio effects are all mixed, making the ASR difficult. We did not test all the languages supported, but for the ones we tested, the accuracy was great. Perhaps better than with Amazon Transcriber.

We liked the good documentation and the examples that allowed us to start testing quickly. We liked, even more, the encouraging and responsive customer support and the information about their special startup program.

What do you dislike about the product?

Amazon Transcriber currently supports, and perhaps will always support more languages (which is important for our startup ) than Deepgram.
This review should not be considered final. We touched, tested and liked Deepgram, but our current development logic dictates us to take a pause with it.

What problems is the product solving and how is that benefiting you?

We build a platform helping to comprehend and learn the language of a playing feature film as your second language. We needed an accurate speech timestamping, to accurately reveal the slightest natural pauses in the speech flow. We realize Deepgram can become our first choice.

Recommendations to others considering the product:

N/A

What do you like best about the product?

Very Good and easy interface. Many different ways to access the tool. Good performance. Better than Vosk and similar to Google. We use mainly use Python as an interface.

What do you dislike about the product?

Couldnt easily find the price for the tool. If I saw quickly it could be cheaper than our current tool I would keep on trying. I would like models that can perform with music in the background

What problems is the product solving and how is that benefiting you?

We do speech-to-text to publicity spots videos in Spanish. We need high-quality text for doing afterward NLP projects. We are trying different tools finding a budget performance equilibrium

What do you like best about the product?

Easy of use - the api is exactly what I need.

What do you dislike about the product?

Lack of reporting on failures. Wasn't obvious until I dug around a bit.

What problems is the product solving and how is that benefiting you?

We need a fast, accurate and reasonably priced STT service. Deepgram covers all that pretty well.

What do you like best about the product?

one of the best things about the deepgram api services, is that it provides voice profiles and can detect if multiple people are speaking in the audio and thus provide with separate speech-to-text for the same.

What do you dislike about the product?

the thing that I dislike about deepgram is it's lack of documentation for integrations with React framework. Apart from these everything is fine in the deepgram and it's services.

What problems is the product solving and how is that benefiting you?

We found a solution to detect inactivity in online meetings, classes, and sessions. We used the MERN stack to construct an extension that generates audio transcriptions and calls the user if he or she is AFK during critical sections of the call/session.

(MERN - MongoDB, Express, React, Node.js)

Recommendations to others considering the product:

One must read all the documentation properly for better understanding and integrating the api.

What do you like best about the product?

The ease of use! The simple but powerful APIs make it so quick and easy to start creating something. Meaning I spend less time worrying about how to handle speech processing and more time using this processing to create a useful product.

Not only were the tools very easy to use but they were also incredibly fast and accurate. I came across no translation issues when using the product despite testing it in noisy and non-optimal conditions. And the results were almost instantaneous meaning that the processing could be run many times with very little impact on run-time performance. The accurate and instantaneous results were one of the best things about Deepgram's tools as I knew I could rely on them in my final product.

Other tools I had looked at were either very restrictive or not very accurate so it was refreshing to find an SDK that gave the flexibility to do whatever I want without compromising on speech and accuracy.

What do you dislike about the product?

Nothing in particular that I disliked, the user experience was great. However, I'd love to see the list of SDKs expanded to more languages in the future to include languages such as Rust and Ruby so I can get easy access to the speech processing no matter which language I am building my project in.

What problems is the product solving and how is that benefiting you?

The first time I used Deepgram was during Hack Cambridge Atlas Hackathon (2022). They were one of the sponsors of the event and demonstrated just how versatile the product was which inspired us to incorporate their tools into our project. Due to the numerous ways that covid has affected our lives, we wanted to make a product that could take advantage of the new changes that are being made. We decided to settle on creating an educational application that could be used in classrooms, utilising speech processing and NLP to enhance the learning experience. One of the most influential changes due to covid was the introduction of tablets and laptops for each student in the classroom, and so we wanted to create an effective way for students to feedback questions to the teacher during the classroom using this new technology.

To test the waters, we build in functionality to transcript what the teacher was saying so students could check back at what was said in case they missed something. This is where we truly got to see the power of the Deepgram APIs as it flawlessly translated the speech into text in only a few lines of code. We then went on to build a functionality where students could ask questions about their devices. One problem we realised was that the teacher may struggle to answer all the questions if they just appeared in a list on their screen. And so we came up with the solution of running NLP algorithms over this text to identify the keywords in the question. This then meant we were able to group similar questions and display them in a graph with similar questions clumped together so teachers could quickly identify topics that students were having similar problems with. One thing that helped with this process was just how fast the API was, this meant that we were able to instantaneously translate and submit questions to the teacher which gave us a similar level of immediate interaction that would be found with vocal communication, but with the extra benefits on NLP and multiple people able to submit questions at once.

Overall this project was a large success and it was only through the use of the Deepgram APIs that we were able to finish our project in time. Using the speech processing tools was simple and yet it was incredibly effective at tackling some of the toughest problems in our project. I enjoyed this experience and look forward to using Deepgram again in the future.

Recommendations to others considering the product:

Use the tutorials as they are very well made, especially if you have not used similar tools before.

What do you like best about the product?

Deepgram offer services in various platform

What do you dislike about the product?

Nothing, everything is peachy
Deepgram is heading in right direction and it offers various platform

What problems is the product solving and how is that benefiting you?

Search location by voice instead of typing

Recommendations to others considering the product:

Yes

What do you like best about the product?

The easy integration of deepgram in projects make it really easy to use

What do you dislike about the product?

It requires credit for each translation which is one downside otherwise its best

What problems is the product solving and how is that benefiting you?

I have made a project which takes the human voice in real time and using those commands it generates website elements

Recommendations to others considering the product:

This is one of the best speech recognition services which is fast and accurate. There are so many applications to this.

What do you like best about the product?

It's ridiculously fast to get set up and going. As in - you sign up, write 5 lines of code, and you're done - kind of fast. And it just works! Its accuracy is really good, it's fast, and it has some fancy extra features to top it all off. And if you have some specialised audio type that most recognition services perform poorly on, they've got you covered too ;)

And for the developers out there, the docs are seriously *great*.

What do you dislike about the product?

I don't know how well Deepgram scales up given it's not a tech giant such as Google, or how accuracy stacks up against the absolute best in the industry, but what I do know is that it performs damn well for 99% of your project needs.

What problems is the product solving and how is that benefiting you?

I first came across Deepgram in a hackathon, where they had sponsored a challenge to utilise their technology in a way to benefit society. Given this broad scope, my team produced some software to automatically transcribe any video on the web into captioned subtitles, and automatically position and move the captions next to the person who was currently speaking.

Part of Deepgram's API allows you to differentiate between individual speakers from within the same audio track, so our next step is to extend our software's capabilities to allow and support multiple concurrent speakers in the same video, correctly placing each speaker's captions next to them.

Recommendations to others considering the product:

For 99% of projects which require good, out-of-the-box speech recognition, Deepgram has you covered. For any specialised cases, I recommend you contact Deepgram directly and see how they can help you.

What do you like best about the product?

An intuitive web interface and a (mostly) good python module got me running quickly, and a quick response time for requests along with real-time transcription let our program feel snappy.

What do you dislike about the product?

We had a couple issues with setting up the speaker identification, which was a bit flaky with multiple sources of audio talking over each other, although it was mostly working as intended.

What problems is the product solving and how is that benefiting you?

We used Deepgram to help with a project in a hackathon, which allowed us to work with the data to provide insights into the topic of the current speech. Realtime transcription allowed us to do this without the traditional method of caching an entire conversation before being able to process it, which simplified the pipeline and resulted in a better project.

Recommendations to others considering the product:

Understand the pricing system, although I believe the prices are fair for the quality

What do you like best about the product?

The wide array of features available - there are so many possible things that you can do, however since I was only a beginner I could just use the basic features until I got comfortable with them, allowing me to move on to more complex features.

What do you dislike about the product?

While only a minor problem, when building our project we wanted to try to run it off Javascript in the browser, but soon found it lacked that functionality and so had to create a new seperate NodeJS server to send it to instead.

What problems is the product solving and how is that benefiting you?

Deepgram gives us the opporunity to create new uses of technology. A project I aim to start soon is using both facial recognition and Deepgram speech recognition to create a 3D avatar, with chat ability.

Recommendations to others considering the product:

Start with something basic to understand how it works, then try all the features provided.