Removal of announcements in the hall over the voice of the reporter (video work)
The other day, I received an order from a video production company for a highly difficult project. Despite the sound of several tens of seconds, we finally delivered it by repeating trials and errors many times due to strict finish requests.
Guidance announcement in the venue is fragile in the report voice at the event venue
The request for audio restoration was the video content to be reported at a certain event. While the reporter is talking about the situation of the venue, the voice of a long announcement that the announcer who manages the inside of the event venue guides the guests in the venue is fragile.
The point is to improve the situation where the voice of the announcement is close to zero by making the best use of the voice of the reporter while making the most of the presence in the hall.
Originally this cannot be adopted as an NG material
It seems that the video cameraman was recording the audio at the same time, but probably he wasn't taking the audio monitor. It seems that it was decided as a material that can be used for this chan without worrying about the announcement in the hall chatting with Bella Bella behind.
The volume of the announcement in the hall was large, and it was recorded with a balance of about 10: 8 with the volume of the reporter who is the leading role. Since the distance to the microphone is different, the sound in the hall can be heard with a little echo. While the reporter is speaking, the voice is drowned out to some extent for a while, but all the time when the reporter is not speaking is completely filled with the voice of the venue announcer.
It can be said that the best way to edit is to use the air audio material of the site sound and replace it with dubbing while looking at the screen, but there is no material that can be mixed (the site sound of rustling that you can not hear the announcement). It's a closing comment that must be included, but it wasn't something that could be retaken.
The hardest part was the discrimination and mitigation of voice components.
The procedure of the editing work that we first performed with RX6 Advanced, which we mainly use for this kind of voice processing.
Silence other than the reporter talking
Using an environmental sound synthesis module called Ambience Match, we capture the "buzzing sound" of the venue from the part of this material that no one is talking about, and synthesize it into the silent part.
While enlarging every part of the spectrum meter and finely pinpointing it, the mitigation process is steadily performed while searching for the voice of the announcement in the hall.
It's easy to itemize, but the work ratio of ①②③ is about 1: 1: 350. To be clear, work while repeating ③ transcendentally and steadily with trials and errors.
Since both the reporter and the announcer are women, the voice is recorded in almost the same band in the voice data, so the cut amount will affect the voice of the reporter you want to utilize.
The first sample proposed version did not satisfy the client in this part. However, due to the high expectations for our technical capabilities, we had the budget and delivery time increased, and after proposing the remaining two takes, we were able to adopt them.
There were almost no cases where it took so much time to edit a few tens of seconds, but there are cases where we can accept it if the budget and delivery conditions are met.
Please feel free to contact us as the first trial is free of charge for such seemingly impossible improvements.
* If quality is prioritized, please fill in the budget of the inquiry form at the upper limit as much as possible. We will adjust the grade according to the budget to some extent. (Because we are always following the risk of rejection when it cannot be improved)
Video audio recording trouble (missing pin microphone recording) Repair case