Reviews from AWS Marketplace
0 AWS reviews
-
5 star0
-
4 star0
-
3 star0
-
2 star0
-
1 star0
External reviews
External reviews are not included in the AWS star rating for the product.
Responsive company with a quality product
What do you like best about the product?
Their on-prem deployment strategy makes my life so much easier than previous ASR products we've self-hosted. Docker images that easily run in the cloud are such a great way to deploy and maintain software. They also make their infrastructure easy to monitor and publish stats to CloudWatch (with minimal configuration).
What do you dislike about the product?
Since this is a relatively new company, it sometimes felt like we tested things for them. There were some transcription discrepancies when switching from our old ASR provider to Deepgram, but they helped us figure those out. These types of nuances between ASR providers are to be expected with any significant switch of ASR systems.
What problems is the product solving and how is that benefiting you?
Our company transcribes data for numerous industries for ingestion into our analytics platform. This data allows us to build a SaaS product that enables our customers to understand what customers are saying about their products and services.
- Leave a Comment |
- Mark review as helpful
Appreciate the fine grained control for real time recognition
What do you like best about the product?
Real time recognition returns final results quickly with punctuation and speaker labels. I can fine tune pause detection and supply keywords to boost on the fly.
I've used this with spf.io to caption and translate live events. Compared to other speech recognition engines, Deepgram gives more fine grained control over when I get finalized results and is simpler to get started with--I just supply my API key and it works!
I've used this with spf.io to caption and translate live events. Compared to other speech recognition engines, Deepgram gives more fine grained control over when I get finalized results and is simpler to get started with--I just supply my API key and it works!
What do you dislike about the product?
I had an initially poor support experience where the real time recognition was not responding with results nor sending any error messages. I contacted support and it took a few days to hear back, so I had to use a different engine for my customers at the time. Later the head of product reached out to me and I was able to give feedback and since then have had much more responsive support.
The vocabulary is not as rich as other engines. Although there are ways to build custom language models, it would be nice if the base model already had a richer vocabulary for the English language.
The vocabulary is not as rich as other engines. Although there are ways to build custom language models, it would be nice if the base model already had a richer vocabulary for the English language.
What problems is the product solving and how is that benefiting you?
My product spf.io is used by customers to make their events accessible in many languages. Deepgram's real time recognition has enabled us to rely on fully automatic transcription in more situations because of the way it returns finalized transcription results with punctuation quickly with reasonable accuracy.
Powerful, accurate, scaleable and blindingly fast!
What do you like best about the product?
Our EdTech platform has tens of thousands of users, when we were using our previous transcription engine (on AWS) scalability was becoming a common issue for us. With Deepgram using the power of GPU's instead of CPU's, this immediately removed our transcription delays as we can scale when needed, and my god, it's fast. We've since been able to process a high number of transcription requests simultaneously, and completing an hour's worth of audio in under 10 seconds is incredible.
I've also got to mention its labeling and learning; this allows us to develop and increase our accuracy over time. Deepgram offer subject matter experts as a service which is great, but we're building it in a way that allows the consumer to make any edits to then feedback in to Deepgram
I've also got to mention its labeling and learning; this allows us to develop and increase our accuracy over time. Deepgram offer subject matter experts as a service which is great, but we're building it in a way that allows the consumer to make any edits to then feedback in to Deepgram
What do you dislike about the product?
So far, we haven't seen a single downside to moving to Deepgram. The functional benefits came to light immediately after swapping and one week of beta testing. One suggestion that I'd have (and this doesn't stop us from using Deepgram) would be to have an assigned Account Manager to speak with, but I should mention that they have an online chat which is also acceptable for us.
What problems is the product solving and how is that benefiting you?
In EdTech, transcription is almost seen as a must-have for SaaS providers. The speed and accuracy is vital to any players in the market. What we had noticed when using previous transcription providers was the considerable delay in receiving the complete transcription, which would quite often lead to our users thinking it's 'broken'. In using Deepgram we've been able to supercharge our transcription overnight.
Deepgram is next-gen transcription
What do you like best about the product?
Deepgram's accuracy is hands-down the best aspect of their service. When using Deepgram we saw over 93% accuracy, and it's only getting better. Deepgram also has an amazing team and support staff.
What do you dislike about the product?
Deepgram is building out more robust compliance standards, which will continue to be needed as they expand into new markets such as finance, healthcare, and SLED.
What problems is the product solving and how is that benefiting you?
Our business relies on accurate transcription to work effectively. Our company does physiographic profiling of users based on things like emails, texts, etc. Our analysis is only as good as the transcript.
Recommendations to others considering the product:
Ask for the Demo. The Deepgram staff is amazing.
Low cost. Good engineering support and APIs works flawlessly
What do you like best about the product?
The response time of APIs and the accuracy.
What do you dislike about the product?
Monologue concatenation can be improved along with numerals conversion.
What problems is the product solving and how is that benefiting you?
Speech to text. Accuracy is awesome.
Recommendations to others considering the product:
Low cost, low latency
Great experience
What do you like best about the product?
Deepgram has a high accuracy with spanish transcriptions
What do you dislike about the product?
In some cases the transcription is not according with the audio, and numbers or urls or mails come in text format. Also the tags of participants and diarization is inconsistent
What problems is the product solving and how is that benefiting you?
To get transcription from audios in spanish language. To simplify our transcription service.
Recommendations to others considering the product:
To have different data to test and verify the accuracy.
Efficient and well-supported ASR
What do you like best about the product?
Deepgram is a modern product that has excellent customer support. The on-prem deployment was effortless. Engineers from the company also assisted us in making some customizations we needed. Deepgram's support has always been available and ready to answer any questions we have had.
What do you dislike about the product?
GPU compute hardware can be expensive in the cloud or on-prem these days.
What problems is the product solving and how is that benefiting you?
Deepgram has allowed us to easily scale our ASR needs and keep our costs down compared to vendors we have used in the past.
Great solution that provides with state-of-the-art speech models. It helps accelerate everything
What do you like best about the product?
The API call for the speech-to-text models returns a really detailed response. The upside is that here you can distinguish what happens in the conversation word by word, speaker by speaker, minute by minute. Also, there's an advantage when checking the punctuated transcript. It's helpful that we get the transcript confidence. The response speed is fantastic because it helps us run things simultaneously while we wait for the response. If you hold stereo-format audios, you get the diarization for two separate channels.
What do you dislike about the product?
Sometimes, the speaker diarization result is not that accurate. Sometimes the response states that there was just one speaker, so we cannot determine who said what. Also, we understand that your Spanish and Portuguese models are not that precise enough, so the downside here would be that our potential clients/clients could wait for some level of precision we cannot give until that solution is available (or even specific new terms or phrases that we get week by week that they would like to be detected, and that won't be there until we retrain models with your service).
What problems is the product solving and how is that benefiting you?
Having the textual transcript of our clients' calls was our most significant pain initially. Deepgram is a fast and reliable solution that has helped us accelerate other processes, so we do not waste time developing something that is already there and focusing on delivering more valuable functionalities to our clients at the same time. Also, a fast response is helpful because it helps deliver faster too. The API response is detailed, so we can play with all the parameters in the response and adapt it to our clients' needs. The new console helps you track the service you've received (e.g., the hours you've consumed), so you can keep your activity followed project by project.
Recommendations to others considering the product:
First, be clear about your intentions when agreeing on a service level. Look for ease of use, hours of data and possibilities for integration. Also, be clear if you need to use a general model or a more context-specific model. Be sure about what parameters inside the API call you will need to use. Validate with audio samples, so you check for models confidence vs. audio quality.
Works better than competing products
What do you like best about the product?
The API is well documented, a joy to work with, and the Deepgram outcomes are great.
What do you dislike about the product?
I don't dislike anything about deepgram. It could always be cheaper :)
What problems is the product solving and how is that benefiting you?
The need to transcribe audio quickly and accurately. Deepgram is faster and more accurate than other services that I have used.
We analyzed ~10 transcription vendors and DG was the highest performing. And have wonderful support.
What do you like best about the product?
Their hands on approach to the customer. The Slack Connect channel they set up with us and handholding during onboarding.
What do you dislike about the product?
Haven't gotten to any dislikes yet. Stay tuned :)
What problems is the product solving and how is that benefiting you?
We are doing real time transcription of Zoom based conferences. We are then overlaying our own data models (NLP) on top of their transcription.
showing 231 - 240