I have something for you to listen to before you start reading. It’s here below. It’s an on-tram announcement from the Manchester tram network.
The small caveat on what I’m about to say below is that what’s above may not be the exact message I hope it to be as they haven’t played that particular message when I’ve travelled with my recorder. What I’m about to say does apply to the message above, but the sample I will continue to try and capture is even worse!
The situation with the tram network announcements is a fairly common one. At the beginning there were dozens of announcements recorded which covered the entire set of circumstances at the time. But the network is a fluid entity as new lines have been added since those initial sessions and new circumstances happening require new announcements. So additional recording sessions are needed and the new prompts are integrated into the system.
During my time as an audio producer I’ve worked on many such systems – both public announcements (I did over a decade producing announcements for the mainline rail network) and many many different IVR systems for many different companies – that have needed prompts adding to the system, sometimes fairly regularly and sometimes not. Covid has meant a lot of clients have changed their opening hours or their terms of service and needed that reflecting in the information on their phone lines for example. So I do feel qualified to say that the prompt above is terrible!
Why is it so bad?
The point of putting this kind of system together is to ‘hide the cracks’ as much as you can and make people think it’s an actual person giving you an actual message. In the prompt I hope you heard the voicing is inconsistent in its speed and it’s tone, and the producer hasn’t matched the EQ or levels.
Hopefully what you’ve listened to is the prompt I want it to be rather than the one I have in reserve. If it’s the one I want it to be it’s the voicing and the production that are at fault, if it’s my reserve prompt I’m afraid I’m mainly picking on the voiceover (although the producer has a hand – we’ll get to that).
Stage 1 for any VO job is the recording
Or so you’d think. Before we get onto the recording we need to consider the recording environment. This is a crucial part of the chain in retaining a consistent tone to the recording. If you walk around your house talking in different rooms you may notice the sound of your voice change from room to room. It’s particularly obvious between your bathroom and the cupboard under the stairs. Step one of maintaining a consistent quality to your audio is to keep your recording environment sounding the same. This is particularly key if your recording space is small as a minor change to a tiny space can be proportionally huge. I did write a little about this in a previous blog but it’s worth mentioning again as the laws of physics don’t change.
So then we get onto the voicing bit
And it’s with this type of job that we prove that being a VO isn’t just reading out loud. You as a VO need to ensure that your read is consistent with what you’ve done before. It’s vitally important to listen back to a previous session/finished prompt and do that again. Take note how well your voice is warmed up, the speed of your read, how formal/smiley/polite you are and how much you’re projecting. You should know the brief from your client as to how friendly different bits of the project should be. The big thing to remember is that this half-sentence you speak will need to fit onto a half-sentence you recorded maybe a decade ago.
The prompt you listened to earlier has failed to note the above. The speed and projection of part of the read is noticeably different to the point where it sounds like she’s throwing half the sentence away. ‘Well, you can go there if you really want to, but why would you want to?’
Now onto the production bit
But let’s stop picking on the VO (I don’t know who it is, by the way. To the best of my knowledge it’s an employee of TFGM) and pay attention to whoever’s produced and prepared the prompts for usage. Earlier on I said that the producer has a hand in the read being off. This is step one in the producer’s role in a production job – making sure the voicing is of sufficient quality to use. This doesn’t just mean ensuring the recording is being done at a suitable sample rate, but it means checking the VO is happy and comfortable in the studio and, when required, QCing the read; ensuring it fits the clients brief and/or matches the previous job if it’s a repeat job or pickups of a previous session. This is kind of presuming the recording session is in-person or a live remote session. But even if it’s a self-record there should be scope for the producer to go back to the VO if necessary and ask for a re-record. I promise I’ll stop picking on the VO now and pick on the producer.
The failure of whoever produced these prompts to notice the inconsistency is a bit of a worry. It indicates not only a lack of attention to detail, but also an inconsistency in production quality. This type of job is a sample of a job best done in a DAW rather than an editor as your DAW will keep your processor settings (compression, EQ, etc) whereas the editor will lose them as soon as their applied to the audio. It is possible to save effects racks/stacks/chains in editors, but it clutters up the workspace if you do that for every individual client/job. Plus there is always going to be the possibility that the tone of the recording will be slightly different and you need to tweak your processing settings to match the original audio.
There are many reasons why the EQ of a read may differ from the original. It may be a change of mic (a producers nightmare!), a slightly different position of addressing the mic, or something as innocuous as different humidity and air pressure. It’s ultimately the producer’s responsibility to ensure the new prompts fit with the old ones. Again for the producer listening back to old prompts is crucial when matching new audio to old. Ensuring the EQ, compression and levels match is the very definition of what the producers job is with this type of work.
This is the perfect type of job to be handed over to AI/TTS in future, so those of us who are involved in the creation of this type of audio system should be doing all that we can to keep this work in our hands. Maintaining a high standard of delivery and production is one of the best ways to keep this in our hands. Allowing such crappy audio as this out into the wild does no-one any favours.
TL:DR This tram announcement is terrible and they should get me to produce them.
Image from UnSplash
To find out more about Rob’s Audio Production services, CLICK HERE >>>