Tag Archives: transcribing

Youtube transcripts – get useless auto-transcripts replaced with a helpful version!

I’ve just been having a late lunch-break and watching a video on YouTube, as you do. One of my many hobbies is crochet, so I decided to watch this video from ‘Girlybunches’ on how to make a ‘hyperbolic crochet brooch – but don’t worry if you don’t care in the slightest about crochet; that’s not what I’m writing about.

I liked what I saw so I looked for the subscribe button. I admit it should have been hard to miss, being bright red an’all, but I wasn’t wearing my glasses! So I started looking through various available buttons and found one called ‘transcript’. Intriguing! Obviously I had to find out more!

What I found out was that machine transcription has a long way to go!

Here is a short sample of what Olivia from Girlybunches actually said:

I just think the half-treble gives you just a little bit more length, which makes it come out a little bit more. Another point is, make this with, erm, now how can I be polite …

Here’s what the machine transcription thought she said:

I just think home trouble Kyushu district bit mornings which makes it come out a little bit more am another point he’s make meese with I’m know how can be polite

Honestly, it’s all like this, I’m not just picking the worst bits!

Here’s another bit.

And magically … and I will put links down below to my video showing how to do these things. You know, you won’t have to worry about not knowing how to do them ‘cause I’ve shown you. And you just do twelve in the loop …

Or, alternatively, from the machine transcriber:

and magically armpit links temple known to my video showing how much do these things time you know you don’t have to you worry about not my cup Stephen itself option you and you just eat well in the …

Now I don’t really know why YouTube provides transcripts – is it to help with SEO? If so then frankly it probably won’t! What comes from the machine transcriber will probably miss most of your essential keywords! Is it to help people with a hearing impairment? Again, if you look at these examples you’ll see that it probably won’t!

Perhaps it’s confused by Olivia’s English accent? OK, let’s find an American video to compare. Here we go, here’s one by CinnaZilla aka the Delta Quadrant. He’s got an American accent and you can find him here. This is what he actually says:

Hey Youtubers, I am starting to create a series of videos on crochet techniques.

Here’s the machine transcription:

pages firsthand and starting too creatine series videos on appreciate techniques

And so it goes on.

So what’s to be done? Well the good news is that you can replace the terrible machine captions with a quality transcription containing the real words that you say. How? Well, it’s quite  a straightforward process. First, you get your friendly neighbourhood transcriptionists (hopefully us here at Penguin Transcription!) to make a transcript of your file.

Then you go to your channel page on Youtube, choose the video you want, click the bar under ‘edit’ and then choose captions. In the captions page you have the opportunity to upload the transcript that you’ve received from us. Then, once it’s uploaded, you click on the box that says ‘sync’ and Youtube will attempt to sync your voice with the uploaded transcript. It takes a little while, but not too long, for that to happen. And then, hey presto, a transcript that’s actually useful!

We’re looking forward to working with you!

Transcription – offshoring, onshoring, in-housing, outsourcing?

Transcription might seem like an obvious thing to outsource and ‘offshore’. After all, ‘it’s only typing isn’t it? It’s not rocket science?’ And yet ‘onshoring’ has been in the news a lot lately, with both positive and negative slants. On the one hand, onshoring could boost the UK economy; on the other hand, the fact that it now ‘costs roughly the same’ to make noodles in China as it does in the UK, according to the recent news story about Symington’s Noodles bringing noodle production back from China to the UK, is an alarming indictment on the state of the UK economy. But it’s not just that companies feel they can now pay even lower wages to UK staff; it is also the rise an rise of wages in China, exchange rate fluctuations and shipping costs too.

So how does transcription fit into this discussion? Well, another recent argument for onshoring has been quality concerns. And this article about IT onshoring suggests a number of other important concerns too: “…time zone challenges, language and other communication issues, high turnover (up to 40% annually in some cases) in offshore locations, intellectual property and security risks (especially in unregulated countries like China), are just some of the unanticipated issues that have plagued offshore development.”

And a number of of those issues could also affect transcription – the obvious one is language. Unless English is a first language then there is no way that someone can provide top quality ‘general’ transcription i.e. interview transcription services and focus group or meeting transcription services. It is possible (though perhaps doubtful) that they can provide equal quality dictated notes, for example, but a conversation – full of idioms, homonyms, a wide variety of different technical terminologies – no.

So … if I’m suggesting you should keep your transcription ‘onshore’ then what about keeping it in-house? Surely keeping it as local as possible will minimise the problems? Well no, not necessarily. And this is where we come to the ‘just typing, not rocket science’ issue. It’s true – it’s not rocket science, but it does require specialist skills, and even if you’re lucky enough to have access to a secretary or PA who can type, that doesn’t mean they can provide fast, accurate, grammatically correct and readable transcripts from an audio file … and all that on top of their regular workload.

I’m sure it will come as no surprise that I am recommending onshoring and outsourcing, since this is the service that we offer here at Penguin Transcription,  but I think you will agree that the arguments are valid.

Language is Evolving – and the Transcriptionist has to evolve with it!

The way language evolves has become a fascination for me since I started transcribing. I started thinking about this again when @Wordnerdsally on Twitter brought my attention to the latest rebuttal, by  Académie Française, of an English word (hashtag), because it’s damaging French language purity. Will they actually stop the Frenchman (or woman, or child) on the street from using the word? I very much doubt it!

The fundamental problem with the whole idea of the Académie Française as a protector of the purity of the French language is that language is not, and never has been, ‘pure’. It changes over time because it’s living and it evolves; people speak it, younger generations love to twist and turn it and make it their own, waves of immigration bring in new words and change old ones, and language just keeps on changing.

A favourite evolution of a word for me, into something more negative than its original meaning, (pejoration, in linguistic terms, so I’m told) is ‘silly’. In Old English, ‘silly’ meant, of all things, blessed! If you were blessed, I suppose you were naturally thought to be innocent, so the word then started to mean innocent. By the time of Middle English (Chaucer’s era) that had evolved into ‘deserving compassion’. Not quite sure how the link worked there but I suppose if you were innocent of a crime and had been accused of it you would deserve compassion – maybe it evolved that way? Anyway, if you needed compassion, that must mean you were weak, right? Well no, probably not, but that seems to have been the thinking then, so ‘silly’ started to mean ‘weak’. ‘From there it was ‘a short step’, says linguist Professor John McWhorter, to it coming to mean ignorant, and from ignorant it evolved into ‘lacking in good sense’, which is one of its meanings today.

This is probably a very over-simplified description of the evolution of silly, not least because it doesn’t only mean ‘lacking in good sense’. It can also mean frivolous and it can be used to describe objects, not just people. However, it’s an indication of the complexity of language and the difficulties inherent in making sense of it! And making sense of language, as it is spoken, and translating that spoken language to something that makes sense on the page, is really what transcription is all about.

The evolution of silly took place over hundreds of years, but some words change much faster than that, especially in spoken English, rather than the more formal English usually found in writing. An obvious recent example is ‘wicked’. When I was a lass ‘wicked’ meant evil, and to some it still does, but most people would hesitate to use it that way because in the younger generations it has come to mean ‘cool’ which of course when I was a lass meant slightly chilly, and not hip (a joint of the body?), groovy (having lines engraved in it?) or just ‘in with the current style’ (from ‘Urban Dictionary‘).

So a good transcriptionist isn’t ‘only’ someone with a fantastic grasp of English spelling and grammar, but someone with their ‘finger on the pulse’ of current spoken English. Fortunately at Penguin Transcription our transcriptionists are not only experienced, but range considerably in age and background. When that the word ‘mardy’ came up in a transcript I was doing I had no idea the word even existed and kept trying to ‘hear’ another word that would make sense to me, but then a younger colleague listened to it and grasped it straight away. (For those who are just approaching ‘middle age’ like me, or older, it means grumpy, surly or miserable apparently!)

Getting the best from a recording for transcription

There are many transcription services available but sometimes an affordable transcription service can seem hard to find. Transcription is not cheap, because it is a lot more involved than copy typing, but that doesn’t mean you can’t find a good deal with a transcription service, and what’s more, by providing good quality recordings you can make the transcription more affordable, as it will take less time to complete.

Here are a few things to consider:

Time Taken to Transcribe

When pricing up your options the most important thing to remember is that it’s just not possible to type as fast as you speak. Even an experienced transcriptionist will be able to average four times as long for a good, clear one-to-one interview – so an hour of recording will take an average of four hours to transcribe. (Industry standards obtained from the Industry Production Standards Guide, published by OBC, Columbus, OH, USA). But a poor quality recording will take much longer. So how can you make sure that your transcript is clear, in order to get an affordable transcription price? Basically, the easier you make the transcription for the transcriptionist, the more likely they are to be able to give you an affordable transcription quote.

Equipment

First of all, use the best transcription equipment you can afford, and make sure it’s right for your needs. This means that for interviews you should have a recorded with an external microphone rather than one built into the recorder, which is only designed to pick up dictation. For focus groups you should ideally have several microphones so that all participants are audible, and for conferences the speakers should have good microphones and there should also be people in the audience with ‘roving’ microphones to take around to any audience members wanting to ask a question.

Environment

Always try to make sure that you are recording in a quiet environment. Open windows can cause big problems unless you have a ‘noise cancelling’ microphone, which many digital ones are these days. So can air conditioning, so if you do have an air conditioning unit in the room try to ensure your speakers are not situated close to it. If conducting interviews by phone, and assuming that you have arranged these in advance (and asked permission to record, of course) then it’s helpful to ask your interviewee to try to make sure they’re in a quiet environment too!

Details

If you are interviewing and you want the names included then it is helpful to spell out your interviewee’s name at the beginning of the recording, before starting the interview, and speak out any information you would like on the transcript header e.g. the date, the job title of your interviewee etc. For conferences a speaker list and also a delegate list, if there will be audience questions, can save the transcriptionist a lot of time in trying to work out names and organisations.

Care with Conversation 

During the interview, unless you need to interrupt in order to take back control of the interview, try not to speak over your interviewee. Often in a normal conversation we say ‘yeah, yeah, yeah’ or ‘right’ or ‘OK’ more to indicate we’re listening than for any other reason. Every time you say that you are likely to be obscuring a much more important word or group of words spoken by your interviewee. And in conferences or panel discussions, if one speaker is giving a talk (i.e. without interruptions, not a discussion) make sure everyone else’s microphones are turned off. I have, in the past, had to mark whole sentences or even paragraphs of a talk as inaudible, because all I could hear were two panel members chatting about their holidays or little Jonny’s operation, and not the speaker!

Format

Most transcriptionists work in a standard format, whether that be tabular, tabbed, interviews shown as initials or full names etc. Again most are happy to work to your specifications, but the standard format might well be cheaper, so think carefully about whether you need something different or not. Find out what the standard format is in advance if it concerns you, and you may be able to adapt it to your needs.

Verbatim?

Finally, give some serious thought to whether or not you need a verbatim transcription. Verbatim transcription includes every repeated word, every ‘um’ and ‘erm’, all those ‘filler’ phrases like ‘you know’ and ‘know what I mean’ that may be repeated a hundred times in one interview, and can also include pauses, coughs, throat clearing etc. if required. Needless to say, this takes longer. If the transcriptionist can filter out all this stuff the transcript is quicker. In my company the cheapest level is what we call ‘intelligent verbatim’ which cuts out all these fillers but leaves the rest exactly as it’s spoken. Different transcriptionists work this differently though, so always check when you’re phoning for your quote. You can find detailed information about our editing levels on our website.

There are, of course, occasions when verbatim is required – depending on your topic it might be required for legal reasons, or you might be studying the language. But if you really don’t need it, don’t end up paying for it!

Price

And finally, remember that the cheapest transcription quote might not be the most affordable one in the end. There is an oft-quoted phrase: if you pay peanuts you get monkeys. Will it really be cost-effective to send your hard-won interviews to the cheapest service if what comes back is gobbledygook and you have to go through the whole thing correcting every other word? How much time will you then waste that could have been spent more productively? Recommendation is always the ideal way to find a service, but if no one you know can recommend a transcription service then look for testimonials. A good company with a strong track record should always be able to provide these. If you’re still not sure, ask questions and base your decision on the quality of the answers. Things you might like to ask are: turnaround time (when will you get the transcripts), confidentiality procedures, whether they have experience in your field, what the standard format is etc.

First Penguin Transcription Client Satisfaction Survey

This is probably something I should have done a long time ago, but I’ve finally set up a client satisfaction survey, which all clients have the opportunity to fill in, and for which they receive a link when I send them their invoice.

So far the feedback has been incredibly positive. The first positive, for me, was that anyone bothered to fill it in at all, but we have had quite a good response rate. It’s only been running a couple of months but her are some of the highlights:

  • 80% of respondents used our website
  • 100% of respondents would use Penguin Transcription again
  • 100% of respondents were very satisfied with the readability of our transcripts
  • 90% were very satisfied and the other 10% satisfied with our transcript quality
  • 90% very satisfied and 10% satisfied with our ability to identify and correctly use technical words
  • 100% of respondents were very satisfied with our ability to distinguish different speakers in recordings with multiple speakers

One client mentioned that he needed to take the transcripts we sent and put them into a different layout with line numbering etc., so I will be contacting him to explain that this is something we can do for him, at no extra cost, to save him a bit of time. We can work with any Microsoft Word-based template that you provide, and can also work in Excel templates, but that will sometimes incur an extra cost as it’s a little more awkward and can take extra time.

Clients who respond are put into a quarterly draw to win an Amazon voucher, and the survey only takes five minutes or so, so no one need feel they can’t spare the time. If we’ve done some work for you in the last couple of months and you didn’t receive a survey request, just let us know and we’ll be happy to send one out to you (although I think and hope that everyone’s had one!)

How to successfully get a conference transcribed

A word on timing

The most important piece of advice I would give as a transcriptionist is that if you’re going to have your conference transcribed you should arrange for completion of the transcription before the conference even takes place! Of course you are going to want to send the transcript (or your interpretation of it) out to your speakers and delegates as soon as possible after the conference takes place, but a conference is a significant chunk of work to transcribe.

Let’s take an example of a conference where the talks (and possible workshops etc.) total 5 hours. Even if you have excellent audio recording equipment and supremely clear speakers, with minimal question and answer sessions or workshops (the point of which I will explain in a moment) the time taken to transcribe is going to be four times as long as the recording – so you’re looking at an absolute minimum length of time taken in this example of 20 hours. Twenty hours of work is probably a minimum of three day’s work for one person, and there’s a very good chance it will take longer.

A good, established transcription company, employing fully trained and competent transcriptionists who are able not just to type but also to proof-read and edit, recognise the correct homophones (words that sound the same but are spelt differently), and punctuate English correctly, is probably going to be booked up for at least the next few days, and if you book in your recording before the conference and agree to send it on a certain date, they will be able to turn it around for you much faster.

Audience questions and participation

Question and answer sessions are often tricky because of the range of different voices involved. This applies to the audience but also to a panel if you are having panel sessions.

For audience sessions, make sure you have ‘roving microphones’ that can be carried around the audience, so that questions are actually audible on the recording. A good conference recording set-up, so that your main speakers can be clearly heard, and individual microphones for each member of the panel are also essential. These may well come with the conference venue but make sure you check this in advance!

What your transcriptionist needs from you

Another very useful tip is to provide the transcriptionist with both a speaker list and a delegate list. Then during the conference ask the Chair to ask all delegates to state their name and position before asking the question. The transcriptionist can then refer back to the delegate list to insert the correct spelling into the transcript. The same applies, of course, to speakers, although they don’t need to state their names if you provide an agenda and they are introduced.

It is also very useful to provide the transcriptionist with any supporting material on the conference that you have available as this will help to establish ‘key words’, words that may be not in common usage but particularly relevant to the topic of the conference. A good transcriptionist will also probably be able to search out most unusual words, but this takes extra time, and if you have already provided material to help, time will be saved.

Audio or video

A videoed conference probably won’t add a huge amount to aid the transcriptionist  although if there are large numbers of slides used then it may be helpful; or it could be equally helpful to simply provide a copy of the slide presentation. The disadvantage of sending video files is their size: sound files are large; video files are huge! The larger your file is, the longer it will take for you to upload it to the internet for the transcriptionist to download it, and the longer it takes the more likely you are to lose the internet connection, which means that you’ll probably have to start all over again!

The choices are to either convert the video to audio before sending, or to simply put the video on a DVD and pop it in the post. We can receive video via our file-sending service, but the issues above do still apply, and for a file as long as a conference it might be quicker to use snail mail!

Organising a conference – get the transcription booked in early!

A lot of conference organisers like to have the key note speech, at the very least, transcribed, and many also like to be able to send delegates copies of all the speeches or publish selected parts of the speeches on their websites.

Some people just publish abstracts sent in by the speakers, and that’s fine, but others want what’s really said on the day to be recorded for posterity, complete with panel sessions, audience queries, workshops and so on.

If you’re one of the latter, you might want to take a look at my article about getting your conference transcribed, which can be found here. I have just realised it’s a little out of date – surely there are no conference centres still recording onto cassette tapes? But the rest of the article is still valid. I will get that last bit updated soon!

The key points are:

  • Book your transcription well in advance
  • Use roving microphones if you are having questions from the audience
  • Provide your transcriptional with as much information as possible, including speaker list, delegate list, keyword list (if possible), agenda and general information about the conference’s content

The first point is probably the most important. A good transcription service is unlikely to be sitting waiting for your call and ready to swing into action when you say, ‘I have 20 hours of conference transcription and I need it back tomorrow please’!

 

Transcribing when a translator is present

We’ve just taken on a big project, bigger in fact than either we or our client realised at first, as he wasn’t sure how many hours of recording he had! All the files have the interviewer, the interviewee and a translator. We only have to transcribe the English – which is a good thing, as between us in the office we only have a handful of French and Spanish, and this is something a bit more exotic!

Should be easy, you’re probably thinking – after all, if the sound file is an hour long you’re probably only transcribing half an hour’s worth! So why, you might very reasonably ask, am I charging this client our standard rate, and not a reduced amount?

Well in fact I have offered a reduced rate if the recordings are really clear and the translator speaks excellent English – but we (my client and I) rather doubt there are any recordings like that! This is inevitable and I am in no way blaming the client, or indeed the translator! Of course in an ideal world, all recordings for transcription would take place in a quiet office space with the windows closed, no air conditioning on (because it can play havoc with the recorder!) and, where a translator is required, the translator speaking immaculate English with no accent, as well as speaking the tongue he’s translating from perfectly.

Unfortunately real life does rather tend to get in the way – and when you’re recording in rural China or in a war zone, or even an oral history at a little railway museum in the UK, all of which are projects we have worked on in the past, the chances of being able to find a nice, quiet office to work in are pretty small. The chances of finding a perfectly bilingual translator are even slighter!

So although we can, in theory, race through the non-English parts and just type the English, in this project the recording quality is quite poor, the translators’ English leaves much to be desired and many of the translators also have strong accents. Also, the nature of conversation between three people means that the discussions are not clearly and neatly divided into English and the other language. Often, while the translator is trying to do his bit in English, the interviewee thinks of something else he wanted to say and interrupts. Sometimes the translator is talking to the interviewee and then quickly throws a few words in English at the interviewer, before replying to the interviewee in the other language. So we really have to keep our ears ‘peeled’ and listen to everything, even though we can only understand half of it!

All these issues mean it’s taking about as long, or sometimes longer, to transcribe as a good quality, all English transcription, so it’s costing about the same. All I can say is it’s a good job that we all like a challenge at Penguin Transcription!

Recording for focus group transcription – some useful tips

If you’re thinking of running some focus groups and having them transcribed, here are a few things to think about from the transcription point of view. If you take these things into account, you should get the best possible recording, which (so long as you use a good quality transcription service) will lead to the best possible transcription! 

  • Check with the participants before the group starts that they do not mind being recorded for later transcription. Do this well in advance as if one person objects you may have to abandon the recording and rely on notes. If you’re conducting the group, making notes at the same time will be very hard!
  • Provide explanations about your research and give background information before switching on the recorder, to save on transcribing time. (With transcribing, time really is money, as we charge per minute of audio.)
  • If you need to have the different speakers identified in the transcription, ask each person to introduce him/herself. This introduction must be more than just a name, as it is needed for the transcriber to get to know that person’s voice. A couple of sentences minimum is needed. If you make it relevant to your group discussion, the participants will feel less self-concious. For example, if interviewing a group of people about their experiences in a particular shopping centre, don’t start by asking how many pets they have and whether they love horses or not; instead, ask them to say their name and roughly how often they visit the centre and what their favourite shop is and why. 
  • Lay down the ground rules to participants before you start e.g. remind them not to talk over each other as this will cause problems for the transcriptionist.
  • Use an external microphone on your recorder. Internal mikes are only suitable for dictation (one voice). Ideally, if you have more than four people, use a series of microphones.
  • Record the group in a quiet place. Background noise can drastically reduce the quality of the recording and increase the time taken to transcribe.
  • Make sure you use a recorder that has a facility for transferring files to a PC
  • Use a file format that is compressed, so that it can be transferred over the internet to your transcription service
  • Check your recorder is recording before you start the group, and check a couple of times at least during the focus group, just in case the battery has run out or any other minor disaster has occurred! 
  • Do not serve food while recording the group as the noises of eating, passing plates, pouring drinks and whispers of ‘pass the crisps, would you?’ will obscure participants’ speech!

For more information please see the focus groups page at Penguin Transcription.

‘Hidden noise’ problems when recording for transcription

Sometimes background noise in a recording is unavoidable, but it should be pretty obvious that it’s going to affect the recording! Examples might be a recording in a train station with lots of announcements and train noise in the background, a very noisy cafe with other conversations going on around you (not to mention coffee machines) or in a room with a bunch of screaming kids. We’ve transcribed all these sorts of recordings – sometimes it’s frustrating but we accept that at times it’s just unavoidable. With noises like this though it comes as no surprise to the researcher when we say, ‘Background noise is a bit of a problem!’

However, there are quite a few ‘hidden’ noises that can also cause problems, and, unless you’re aware of the possibility of them being a problem, it’s likely that you won’t notice them until it’s too late. Some of these are avoidable if you are prepared in advance – but some will fall into the ‘just have to live with it’ category, as above! However, hopefully the ‘hidden noises’ below will provide you with a few extra things to look out for before starting your recording.

  1. Silent ring. A mobile phone, even if set to silent ring, can interfere with the recorder, so that for the period that it’s vibrating or silent ringing the recording is inaudible. If you need to have mobile phones set to silent, we suggest you place them as far as possible from the recorder/microphone.
  2. Taking notes. When you take notes, you will hardly hear the sound of your pen shuffling across the paper, but the recorder will, if you are writing next to it! We have had recordings sent in where the speech is actually inaudible because the interviewee is sitting a bit away from the recorder and the interviewer is sitting almost on it and busily scribbling notes! The simple solution is to make sure the recorder is closer to the interviewee and to make sure that, if you need to take notes, the pad is not very close to the recorder/microphone.
  3. Shuffling papers. Similar to above – if you have paperwork, or anything else for that matter, such as a handbag, right next to the recorder, then shuffling or rustling noises can sound very loud.
  4. Wobbly recorder. Not a common problem, but we have one client who always seems to have it – if the recorder isn’t lying flat and the table or whatever the recorder is sitting on wobbles a bit, the recorder will rock and the sound of it moving will be considerably louder than the sound of the people speaking!
  5. Air conditioning. Air conditioning, while not sounding especially loud in the room, can interfere with some recorders and make the recording useless. DO take a short sample recording if you’re in a room with air conditioning, while the air conditioning is running, and make sure it’s OK. If it isn’t and you can’t change rooms, open the window if possible! The sound of traffic/building works etc. outside isn’t ideal either, and may cause some inaudible sections, but that’s better than an unusable recording!