3 types of transcription

If you’ve just started looking for transcription services online, you will immediately confront a horde of providers and transcription types. Although the specifics between different transcription services matter, it can be hard to see the forest for the trees. 

In reality, there are three main types of transcription service available online that you need to understand: ASR, Human and Industry specialist transcription services. Knowing the differences between these three services — the difference in price, accuracy and process — will help you make an informed decision as to your needs.

What’s important to keep in mind is the transcript you want at the end. There’s no point hiring a transcription service that doesn’t produce a workable end product. Our goal here is to help you make the right decision from the start.

1. ASR: Automatic Speech Recognition — speech-to-text software

ASR, or Automatic Speech Recognition, is artificial intelligence software that converts human speech into text. What ASR allows is for fast, cheap transcription. ASR converts audio and visual files into soundwaves which are then fed through a machine. The machine then filters out the background noise, converts the sounds into ASCII characters and you have your transcript.

Sounds great, right?

Well, sometimes. See, ASR is good if price is all you care about, and you’re willing to sacrifice accuracy and potentially some of your own time. ASR quality varies dramatically but with 85% accuracy at the top end, with pristine audio quality. Background noise, cross-talk or any other factors complicating the process can quickly drop that figure into the single digits. Even though ASR can deliver a transcript in seconds, if you need a workable transcript, you’ll always have to spend time editing. If your audio quality is poor (e.g. an interview recorded over the phone), ASR may return nearly incomprehensible results. 

You also get fewer options when it comes to the transcript delivered. ASR will seek to transcribe every little detail of the conversation, right down to the awkward stutters and the ‘umms’ and ‘errs’ you utter. Even if done accurately, this can lead to a jumbled, confusing transcript that is hard to read. 

Ultimately, audio quality and outcome are the two things you need to consider when looking at ASR. You should also think about the number of people involved. ASR will struggle with speaker recognition, and the more people you have talking, the more cross-talk there will be — degrading the quality. Luckily ASR is cheap, and some services are free. If you experiment with your own audio, you can develop benchmarks that will help you understand the kinds of audio files that benefit from ASR transcription.    

Price point: Free – £0.10 per minute 

Options: Speaker identification, time stamps

Best use case: Non-professional audio notes with high audio quality 

If accuracy is not as important as price, then ASR software is perfect for you. But if you require accuracy, simplicity and options, then ASR is not the right choice.

2. Human Transcription

Human transcription services really need no explanation. Hiring a transcription service provider with professional transcribers will naturally give you the best result. Transcription that uses transcribers is usually more expensive, with companies charging £0.50-£2.00 per minute, whilst with ASR you could be looking at £0.07-£0.10 per minute. Human transcription is more expensive — but you get what you pay for. 

Transcribers are accurate — they’re able to parse some of the worst audio quality and even accurately identify speakers. If you want every little detail, human transcription services can deliver that as well with  a number of options for this kind of transcription service. In reality, even if you want an ‘exact’ transcript, you don’t actually want every little noise on the recording written down on paper. False starts, stutters, ‘ums’ and ‘errs’ — all of these distract from the readability of your transcript. Standard human transcription services will remove these irrelevant utterances, delivering a far cleaner and easier to read transcript. These transcripts are normally called ‘verbatim’, ‘intelligent verbatim’ or ‘clean verbatim’. 

However, what is perhaps more helpful are transcripts that take the editing one step further, removing all off-topic chit-chat and paraphrasing long, monotonous passages. The edited services, often called ‘Detailed Notes’ are also cheaper, and are a great option to explore. They prove that sometimes the most valuable transcript is a result of what’s not included rather than an exact record of speech regardless of relevance.

Fundamentally, human transcription services give you options and accuracy, but they are more expensive and slower with turnaround times ranging from 12 hours to 7 days depending upon your urgency and budget.

Price point: £0.50 – £2.50 per minute. However, the cheapest end will likely come with corners cut or security concerns.

Options: Speaker identification, time stamps, different levels of detail, turnaround flexibility 

Best use case: Professional and corporate settings, and any instance where accuracy and detail matter, or audio quality is poor 

3. Industry Specialisation 

The third and final transcription service is Industry Specialised services. Industry specialists, are usually human transcription services that focus on specific industry verticals, rather than ad hoc transcription buyers. Although these companies do generally engage with everyday transcription buyers, their focus is on providing bespoke outcomes to businesses. This might be legal transcription, market research, or medical transcription services — all are common and industry specific.

Industry specialists bring two big things to the table. First, they understand the industry. That means they will be able to breeze through jargon with ease and more easily understand the context of what is being said — delivering more concise edits for Verbatim and Detailed Notes transcripts. Second, they are generally interested in engaging in longer-term partnerships — offering discounts for repeat business. This is great for professionals like market researchers who have high-volumes of transcript requirements.   

If you have a specific use for transcription, then it may be beneficial to find an industry specific partnership who can help you in the long term. Having a dedicated partnership rather than using an ad hoc or transactional relationship can be the best way forward. Finding a company that can be trusted to deliver accurate and timely work means that you don’t have to worry about errors in pieces that rely on accuracy. 

They can also provide many of the add-on benefits that we discussed about human transcriptions. Annotations, time stamps and speaker identification are all useful when accuracy and detailed knowledge are key.

What’s more, having a dedicated partnership usually comes with benefits, like quicker turnarounds and even price breaks on work. This means that having an industry specialist could benefit your business even further. Something to consider if you rely heavily on transcription services.

Price points: £1.50 – £2.50 per minute. The same range as premium human transcription services

Options: Fully bespoke services allowing you to get anything you need

Best use cases: Industry specific professionals who need on-going transcription support 

Get the exact transcript you need

Choosing the best transcription service for you all depends on how accurate you need your transcript to be. ASR can be good in some cases, if you’re producing transcription for a series of voice memos, or if you’re recording one side of an interview. But the higher number of speakers there are and the more background noise there is means that your best choice is almost certainly a human transcription service.

Whilst ASR is improving all the time, human transcription is still the best in terms of accuracy. The expense is worth it when you consider the time that you’ll spend going through AI generated transcriptions and correcting its errors, meaning that human transcription is still the best value prospect. 

Whether it’s an industry specialist, or an ad hoc agency, human transcription remains the most accurate, most reliable form of transcription service.