You want the best transcript you can find. But, there are a lot of options to sift through. Different services offer different price points, and all are seemingly ready to deliver a flawless outcome. How do you decide? This guide is here to help.
To start with, there is a big divide in the transcription services market of which you need to be aware. Speech to text software, also known as automatic speech recognition (ASR) programs, and human transcription services. We will cover both, but a little context will go a long way.
Speech to Text Software
Automatic services use algorithms to sift through recordings and attempt to deliver a verbatim transcript. The benefit of this approach is that it is cheap and nearly instantaneous, sometimes capable of real-time transcriptions. They often operate on subscription models, or with per minute costs, averaging around £0.07 — although some are free. The problem is a lack of quality control.
All speech to text software struggles to achieve greater than an 80% accuracy rate even under ideal circumstances. With complications (poor audio quality, multiple speakers, fast talkers, overlapping speech, thick accents, background noise etc.) accuracy can take a sharp decline. If more than one of these factors is present, you can end up with completely unintelligible outcomes.
Your ability to effectively use automatic speech recognition software will be dependent on your willingness to spend time editing the transcript and the quality of your audio file. If you go to the effort beforehand to create an immaculate recording, ASR might serve you well. The low costs make it easy to experiment. But, they do not produce transcripts that you can count on.
Human Transcription Services
Human transcription services have a much greater tolerance for challenging audio. The human brain is still far better at linguistics understanding than the best machines on the market. Human transcription services also generally make proofreading guarantees. Effectively, you can count on a quality outcome.
You can also choose between different outcomes. Most commonly, this means editing out stutters, repetitions and the ‘umming’/‘ahhing’ that can make reading natural speech difficult (intelligent verbatim), or the inclusion of all of these specifics in addition to added notes on pauses, tone and laugher (verbatim). Lastly, some transcription services offer ‘notes’ or ‘summary’ options that forego detail to summarise the major points of a recording — allowing for the quick digestion of information.
The trade-off is cost and turnaround time. Most human transcription services hover around £2 per minute of recorded audio and a standard turnaround time of 24 hours. Some are cheaper, but lower starting prices are often accompanied by additional fees that increase the effective end price. This can include additional costs based on the number of speakers, audio quality, timestamps, accents or turnaround times. Some services do effectively cut costs and pass on savings to consumers by using offshore labour. However, this comes with security risks that you should consider if having sensitive material transcribed. On average, it will take a trained transcriptionist 1 hour to transcribe 15 minutes of audio. Think about that when assessing prices and turnaround time.
But, you still need to know which service to pick. Here is our list of the best transcription services in 2019, both human-based and ASR software. Each has a speciality. Your specifics will impact which is best for you. In no particular order, let's get started!
Rev: A Transcription Service Giant
Rev is a transcription service company based in San Francisco, California. They have become one of the largest players in the transcription service space. They provide transcriptions, captions, foreign subtitles and translations. Foreign subtitles cost $3-7 per minute, and translation $0.10 per word. Like most services, they use a distributed network of transcriptionists. They have low rates and simple pricing, but fewer options.
Their transcription and captioning services are $1 per minute of recorded audio. This is a flat fee and is not impacted by audio quality, the number of speakers or turnaround times. It costs an additional $0.25 per minute for timestamps and $0.25 for a verbatim transcription. Their standard service is ‘intelligent verbatim’ — although without grammatical fixes, simply the removal of filler words. They do not have a summary option.
They guarantee 99% accuracy for audio files that are clearly audible and are willing to work with clients to correct discrepancies. They offer a standard turnaround time of 12 hours, although that can be longer for files over 30 minutes. But, for all files, they state the possibility of quicker delivery. They use TLS 1.2 encryption to secure your data.
Scribie: Manual and Automated OptionsScribie is another California based transcription company. They are somewhat unique in offering both automatic speech to text software solutions and human transcription services. Accordingly, they have wide price ranges.
The cost of their automated service is fixed — £0.10 per minute. Like most automatic services, the variable is transcript quality. For manual transcription services, Scribie offers a lot of options. If you are willing to wait five days for a transcript, their starting rate is $0.60. Their regular fee is $1.20 (36 hours turnaround) and a ‘rush job’ (12 hours) starts at $2.40 per minute. They come with timestamps by default.
There are, however, then a number of possible additional fees. They will levy a $0.50 fee for non-American accents or background noise. If you want a strict verbatim transcript or other custom formatting that will cost $0.50 - $1.00 extra, per minute. Poor audio quality can cost an extra $0.50 to $2.00. Compared to some services, you might end up spending more, but there is a significant amount of flexibility. If you have ‘clean’ audio, Scribie is a good place to turn.
GMR Transcription: The ‘American’ Brand
GMR is a US-based transcription company, founded in 2004. They primarily offer human transcription services. However, they offer editing, proofreading, and translation services, along with a speech to text app called Easy2Transcribe that they support for free. For their human service, they make a 99% accuracy guarantee.
Their human transcription services operate on a number of different pricing options. Up to two speakers is standard, with their ‘discount rates’ (3-4 week turnaround) starting at $1.25 per minute of recorded audio. Their ‘standard’ service is $1.50 for a 3-5 days turnaround. Fees continue to rise for increasingly quick delivery until you reach £3.50 per minute for same day turnaround. Difficult audio quality triggers an additional $0.50 per minute of audio. More than two speakers will cost you between $0.25 and $0.50 depending on the turnaround. Time codes, detailed verbatim or corrected grammar will also cost you an extra $0.50 each.
GMR fees are slightly higher than other companies. However, they offer access to transcribers located solely in the US. They stress that this adds to the quality of output, security of your data and the delivery of customer service. They use SSL encryption and maintain a private network only accessible by GMR transcription web applications. Fundamentally, their pricing scheme makes their services better suited for projects with flexible delivery times.
TranscribeMe: High-Volume Specialists
TranscribeMe is able to get large jobs done quickly. Like a few other services, they offer both automated and human-based capabilities, ranging in starting prices of $0.10 to $2 per minute of recorded audio.
Their most popular service is called ‘rough draft’. This is a somewhat unique service that trades accuracy for speed and cost savings. These transcripts are not edited, and come with an ‘approximately 98%’ accuracy guarantee. However, this gets you a starting rate of £0.79 per minute of audio for guaranteed 1-day delivery, regardless of the project size.
For long projects, TranscribeMe uses a ‘micro-task’ service in which they disperse your file across a number of transcribers, allowing them to make much more specific turnaround guarantees, regardless of the project size. For their ‘rough’ draft service, this how they can make their ‘included’ 24-hour guarantee. For all other transcription types, this micro-task service will cost extra, and the guaranteed turnaround time is extended to 48-hours.
Their fees then scale up on a large number of variables, including the technical nature of the recording, the number of speakers, timestamps and the type of transcript. They do, however, make a 100% accuracy guarantee for their strict verbatim transcripts. TranscribeMe provides a handy ‘build your own quote’ engine in which you can input all of your specifics.
TypeOut: UK Based Customer Service Experts
TypeOut is a UK based company founded in 2008 to help inventory clerks in the letting industry. They have since expanded to service clients across the world. They pride themselves on customer service, claiming to answer all incoming phone calls in person. All of their management and transcriptionists are based in the UK, making a guarantee that they never offshore work. TypeOut will match clients with transcriptionists with industry experience and use high standard encryption to store and share material.
This high attention to quality and security is complemented by relatively cheap starting fees. Their standard turnaround is 2-4 days, ranging in price from £1.20 per audio minute to £1.85 depending on the number of speakers. You can get a price break starting as low as £1.05 per minute if you are willing to wait 3-5 days for your transcript. Their ‘express service’ (24-48 hours) ranges from £1.35 to £2.05 per minute of audio. Like most services, they levy additional fees for verbatim transcripts, poor audio quality and storing accents. There are also fees for the identification of speakers. However, all of these fees are less than £0.40 apiece.
Transcription Panda: Low Prices For Slow Delivery
Transcription Panda is a newcomer, founded in 2016, they offer quality human-based transcription services. Primarily an American company, they have grown their client base across the globe with a standard service starting at $0.95 per minute of audio. However, that figure is for a turnaround time of 7-10 business days. For a 24 hours service, their fees start at $2.20.
Their low price point for delayed turnaround makes them an ideal company if you can wait for your transcript, but also want to access the quality of human-based transcription services. Like most human-based services, they charge extra for verbatim transcripts, timestamps and difficult audio.
Go Transcript: Bulk-Order Discounts
Founded in 2005, Go Transcript is a Scottish company with over 20,000 transcriptionists ready to start work on your audio files. They deliver transcriptions, captions, subtitles and translations. They offer low rates and fast turnarounds — starting at £0.56 per minute of audio and the possibilities of getting your transcript in 6 hours. This low starting rate is only available, however, for bulk order. They start giving discounts on orders over 180 minutes of audio, and continue to cut prices at several bands up to 720 minutes. This makes them a useful option for those with large orders.
Go Transcript also gives you cost options around turnaround times, with fees of £1.96 per minute for 6-12 hours, all of the way down to £0.71 for 5 days. Their service, however, costs more for low-quality audio and more than 2 speakers.
Otter: Fully Automated Freemium
Otter is a fully automated transcription service. They deliver an iOS and Android supported app that can be used on your phone to transform recordings into transcripts. The app itself can be used to make recordings and is capable of creating real-time captions that can be exported in a number of formats.
Unlike a lot of services, Otter will give you 600 minutes of free transcription time per month. This can be upgraded to 6,000 minutes for $9.99/month. They also deliver enterprise flexible solutions. For individuals, however, Otter is a great tool to use in order to experiment with speech to text software. Its free option will allow you to understand the basics of what types of recordings will work for automatic speech recognition (ASR) software, and which won't.
Temi: Simple and Automatic
Temi delivers a clean and simple automated transcription service. They have no subscriptions, no additional charges and no minimum. Temi is simply $0.10 per minute of audio. Transcripts are delivered back in minutes. They provide access to one free trial off any length. For clean audio (no background noise, clear speakers and minimal accents) they promise to deliver 90%-95% accuracy. However, they are upfront that results may be unusable with heavy background noise, crosstalk or strong accents.
Trint: Automatic Options with Security Guaranteed
Trint is an automatic transcription service with a lot of pricing options. You can pay by the hour or pay per month. This starts at £13.20/hour or £36/month for 3 hours, and 12/hour beyond that. You can also upgrade to their ‘supercharged plan’ for £100 per month, delivering 10 hours and £10/hour after that. They have a free trial, so start there.
Trint is security focused. They guarantee that no person will ever see your file and offer ISO/IEC 27001 accredited environments, along with enterprise customer level support. This can include team account administration, centralised billing and volume discounts.
How to Know Which Transcription Service is Right For You
The big divide is deciding if you want to use human-based services or automated speech to text software. If you need quality transcription service, you have to pick human-based options. If you want a cheap transcription service, or a quick outcome, automated services might work for you.
Certain levels of audio quality, however, can rule out automated services, no matter how many errors you are willing to tolerate. Eventually, it gets to the point where the transcript will be unintelligible. You will almost always need to put some of your own time in editing an ASR transcript. Really, you need to think about automated services on a recording by recording basis. Experiment with some of the free trials on offer and get a feel for what types of recordings work with automated services.
For human-based transcription services, the decision between companies mostly has to do with cost, security and the complexity of the pricing model. Many of the ‘all in’ and simple pricing models cost slightly more than those with a lot of options. But, that is only if you don’t access those options. The ad hoc additions can really add up if you want to access to all of those choices.
Think about timestamps, how many speakers there are, accents and audio quality. If you have a lot of audio, look for a service like Go-Transcript that will offer you a discount. If you need a quick turnaround guarantee on a long audio file, you need to look at services like TranscribeMe that deliver that capability. Equally, if you can wait for your transcript, companies like Transcription Panda or GMR will offer you a better deal for a slow turnaround. If you don’t want to think about it, go for a company like Rev that just has one option. If transcription service security is a big deal, look for companies like TypeOut or GMR that make explicit guarantees that they do not use offshore labour.
The last factor with human-based services is strict verbatim and summary options. Most of the human-based services deliver an intelligent verbatim option as default. But, they don’t all offer the other transcription service choices. Think about this if it is something you are interested in. Fundamentally, however, you should explore the pricing guides on company websites. Then call the one you like best and ask for a quote. Happy hunting!