Amazon Transcribe Reviews

We are using Amazon Transcribe to convert voice to text. For example, we communicate over the phone, record the call, and then convert the conversation into text messages. We use AWS Bedrock LLM models to analyze the conversation, understand the reason for it, and determine any questions asked and answers given. In another project for an insurance company, we use Amazon Transcribe to convert conversation recordings into text. This text is then shared with another team for processing, allowing us to create more advertisements for the user and increase profit. It's about recording the user-agent conversation related to insurance claims and processing it to make targeted advertisements.

The valuable aspect of Amazon Transcribe is its ability to perform speech recognition and convert it into text. It's highly compatible with a serverless environment, making it easy to trigger the service and get results. Although no specific features handle diverse accents or dialects effectively, the scalability and ease of use are notable. It provides the best results for our needs, is highly scalable, and easy to manage. The service also benefits from cost savings, being a pay-as-you-go model with very reasonable pricing for audio transcription at $0.004 per second.

There is a need to improve the processing of background noise. Sometimes, surrounding sounds are recorded and Amazon Transcribe does not process these well, creating clutter. Adding functionality to remove background noise would be beneficial.

I have been using Amazon Transcribe for two years.

There have been no performance issues with Amazon Transcribe.

Amazon Transcribe is very easy to scale. Using DynamoDB and Lambda within AWS makes the process seamless and efficient.

There has been no need to contact technical support since everything is available in AWS documentation.

Neutral

We have configured some Lambda functions using EventBridge, which triggers the Lambda at specific times to fetch and process voice messages from an S3 bucket through Amazon Transcribe. It takes approximately twenty to thirty minutes to set up.

The pay-as-you-go model is cost-effective, with pricing for audio transcription around $0.004 per second.

It totally depends on the requirement, so I have no specific advice. I rate Amazon Transcribe as eight out of ten.

Public Cloud

Amazon Web Services (AWS)

I am a software developer, and I develop medical applications that transcribe appointments between doctors and patients. Afterward, the transcription is processed to create a report based on specific medical structures. I used AWS Transcribe for this transcription service, which I utilize daily. This was one of my applications. 

Another application is related to an educational platform similar to Udemy or Coursera, where video content is uploaded. My client wanted a platform where the video and text description are uploaded, and actions are automatically written by the system. I implemented AWS Transcribe to get the transcription and processed it with artificial intelligence models, such as OpenAI, to obtain the processed text for each video.

The feature I utilized the most was transcription. Initially, I considered exploring real-time transcription, but it was not required for my clients, who needed asynchronous transcription. 

Once the appointment was uploaded, the audio or video could be transcribed. The transcription capability was a core component of business ideas, which was not possible to implement three or four years ago. With AWS Transcribe, transcriptions of video and audio can be processed with other LLMs, like Bedrock or OpenAI ChatGPT, to create custom solutions based on the content.

Several AWS products are originally built in English and not in other languages. There is room for improvement in creating more products in Spanish for Spanish-speaking countries. For instance, the AWS Transcribe medical feature is not available in Spanish. 

Additionally, more detailed instructions for implementing streaming methods would be beneficial. Although the service can perform real-time transcriptions, the implementation process was challenging to understand. Improving educational resources, such as code and examples for real-time transcription, would be advantageous.

I have been using it for already one year.

I have not encountered any problems, so I would rate it a ten out of ten for stability. I tried different providers in search of a cheaper option, and the stability of other providers is lower.

There could be an improvement for scalability which I would rate as a five or six out of ten. For instance, AWS Transcribe offers features for understanding medical content in English, however, this is not available in Spanish. There is room for improvement in adaptability to different contexts.

The technical support can be associated with the technical documentation, which I would also rate ten out of ten. It includes many examples for the standard implementation of Transcribe.

Positive

I have been using a Whisper model from OpenAI. That is the cheaper option I mentioned earlier.

I would rate it between nine to ten out of ten for setup. It is fairly easy to implement due to the SDK, a code package for integrating the service into your code. The process is straightforward, thanks to extensive documentation.

I handle everything myself.

There have been significant efficiency gains. When attempting to implement a cheaper version for transcription, it took several days. 

Direct implementation with the SDK code and deployment with CloudFormation was straightforward. This was profitable in terms of effectiveness and helped me present a minimum viable product to potential clients, convincing them to hire me.

The Whisper model from OpenAI was the only alternative I tried. However, Transcribe was indirectly implemented because of its integration with AWS services. Transcribe automatically records the transcription in S3 buckets, eliminating the need for setting permissions, making the configuration process more secure as credentials are not shared.

I would give the service an overall rating of eight out of ten. There is room for improvement for different languages. 

Additionally, improvements in technical implementation for capabilities like streaming are needed. 

Public Cloud

Amazon Web Services (AWS)

I use Amazon Transcribe primarily for Skype and Webex sessions for Cloudera. I have an application that I use mainly for conducting classes at the start of a Sprint. 

I always record these classes, and sometimes students request transcripts of the lectures. I use Transcribe to convert my speech to text and deliver it as both audio and video to my students. 

Additionally, I attend recorded meetings and use Transcribe to convert them to text and summarize the highlights. This is especially helpful when I miss a meeting.

The speech-to-text functionality is the major feature that I find most valuable, and it has been very useful for me. The results I get with Transcribe are near-perfect—over 99% better than what I have experienced before. This high level of accuracy has improved my efficiency, particularly in delivering transcriptions of my classes to my students. This improvement has enhanced both my financial gain and work efficiency.

There are not many limitations, just my own limitations at times—such as uploading large video sets, which can take some time. I might need a network optimizer for that use case. 

Amazon S3 offers something like uploading parts, where a large file is divided into smaller parts, uploaded faster, and later reassembled. A similar feature in Transcribe would really help, making it easier to upload large file sets without spending extra time.

I have been working with Amazon Prime Catalyst for about a three-year period. Though it is not my everyday service, it is something I have used.

Usually, any stability issues stem from my end, potentially a network issue.

So far, I have not had issues reaching out to support. My interactions with Transcribe have always been similar.

Positive

Previously, I used Google Voice to Text, and occasionally, Amazon Transcribe, but that was a while ago.

The setup is very easy.

I implemented the solution myself.

My billing for Transcribe varies. As mentioned, the cost for Transcribe is usually very low. I would need to check my records for exact amounts, but it is typically minimal.

I rate Amazon Transcribe nine out of ten. It is an awesome tool, although not everything is perfect. Sometimes it is not a ten, however it deserves a nine or nine and a half rating. 

I am looking at different solutions for accessibility.

Amazon Transcribe helps me not to fall behind in a meeting and not know what's going on. Also, I have the transcript at the end to help me figure out what was said during the meeting.

I definitely like the ability to fix the things that are wrong in the transcript. I think it does allow you to put in words that you use all the time, to create your own lexicon, so that it doesn't keep putting the wrong word in every single time.

There's such a huge hole that's missing for people with disabilities, for example, for people who just cannot see or cannot hear at all. They're not just cognitively impaired. I would love to see Amazon Transcribe have its own section or its own page about how to make adjustments if you're using it for accessibility.

It would be great to have closed captions that are in real-time.

I've only been using it just as it is on AWS, and it has been stable in this context.

I'm still trying to decide which one to go with, but I also tested Amazon Polly. I think the price on the standard is better for Amazon Transcribe than it is for Amazon Polly.

With my limited experience, I would rate it at seven on a scale from one to ten.

Lately, I've been using AWS Transcribe for transcriptions. I do have some videos that are a client requirement that convert meetings and videos into audio so that audio can be used as an input factor of the transcript can get transcribed.

AWS Transcribe is the most useful feature for us right now. It allows us to easily transcribe meetings and videos.

Overall, the solution has been very useful.

We don't run into any issues with bugs or glitches.

The customization is good.

We haven't had any security issues.

The solution offers very nice dashboards.

The initial setup is very easy. They offer good documentation, which helps the process along.

Currently, I can't think of any features that are lacking. It does everything we need it to right now.

The UX and UI could be improved on the AWS console.

I've been using the solution for quite a long time. I started with it three years ago, now I think it's been more than two or three years at this point.

The solution is quite stable. I haven't had issues with bugs or glitches. It seems to be reliable.

I pretty much just use the solution for my own personal learning only. I'm currently working on ten projects. There is also a client that I'm working with. They will be buying their own AWS console and they'll be using it. I might just scale them out with it. 

Therefore, I haven't tried scaling yet. I may be trying to scale in a month or so.

I do plan to increase usage and take on more projects in the future.

I really haven't dealt too much with technical support. Mostly, I've been able to use their existing documentation to get the help I need if I run into any issues.

The initial setup is not complex. We found the implementation to be very straightforward and simple.

The product has very good documentation which I was able to access. It's on AWS and it helps us with aspects of the setup and troubleshooting.

The solution requires very little maintenance.

I handled the implementation myself. I didn't need the assistance of a reseller or consultant.

I looked into various cloud services, and I found the pricing to be all pretty much the same AWS is pretty standard industry pricing.

I did look at Googe and Azure before ultimately choosing AWS which seemed to fit my requirements a bit better.

I'm an AWS partner.

I'm using the latest version of the solution at this time.

I just come to our AWS console and I can look for new releases. I do some quick demo's and I try to understand what the releases are all about. I might keep findings for two or three months.

I'd recommend the solution to others. Their technology is top-notch.

I'd rate the solution at an eight out of ten. Although when I first started using it, I wasn't the biggest fan, I've found it's met all of my requirements nicely.

Public Cloud