Enhancing vintage radio recordings

[Pfraser]

Yesterday on Hidden Treasures, I tried to enjoy the radio adaptation of John Wyndham's Trouble with Lichen. I threw in the towel after about an hour, due to the amount of distortion on much of the dialogue.

In fairness, there was a warning of issues with the sound before the programme. I also note that the recording came from a listener's collection, not a BBC archive. It's just a shame that nothing was done to reduce the distortion. Tools for this purpose are available.

On the other hand: should vintage recordings be left strictly alone once any playability issues have been resolved? I do find a certain charm in the spectral limitations of old programmes, such as Round The Horne. I don't mind a moderate level of distortion either.

How did the technical standards at the BBC in times gone by compare with their current ones? Am I in danger of expecting too much from vintage fare?!

[Chrispy57]

Hi - I suspect this is the key to the poor quality, rather than BBC standards of old - "the recording came from a listener's collection". The BBC could be criticised for failing to reduce the distortion if that was the case, running a segment through one of your linked tools may prove the point.

Cheers
Chris

[knobtwiddler]

There is a thread from 3 years back, that links to a site where an ex-Beeb engineer writes about the evolution of recording media during the time he worked there. If you didn't catch it, here is a link:

The Rise and Fall of Audio Tape at the BBC, by Ian Astbury

By the late '80s, the Sony DTC-1000 had become ubiquitous at the BBC. By the early '90s, solid state recorders had taken root. With either medium, you'd have to do something wrong for there to be audible distortion (even the earliest DAT machines could manage around -90dB THD relative to 0dBFS). To give you an idea of how good this is, bear in mind that 2nd harmonic can be into the -20 to -30dB range with many vinyl recordings - and they are considered quite listenable (a modern DJ cartridge can show as much as 3% THD).

Due to ambient noise + hiss / distortion that's inherent to the room / audience / recording chain, I am rather cynical of anyone who claims to be able to recognise the recording medium in many instances, as the artefact that's inherent to the recording will swamp that of a decent machine.

Back in the days of home cassette (or consumer 4-track reel), few people bothered or knew how to maintain their machines. Heads can get stubborn residue stuck to them that even experienced engineers struggle to shift. I have a live recording taken from the local BBC station in 1987, which I made to MAX-G tape with Dolby C. Had I taken a DAT in parallel from the same tuner, I very much doubt anyone would be able to discern between the two, simply because of the crowd noise and variation in SQ of the in-house engineers at the venue. Decent FM tuners could nudge 80dB SNR apparently, which is 8dB better than Dolby C on metal tape! Unless someone at the BBC has a reel of the performance, I doubt they will have a better recording than my cassette!

NB - My pal uses Izotope at his studio. It's amazing at isolating instruments / voices (and creating legal concerns with the results...). Vocals can sound a little unnatural, though - a bit 'formanty'. I would imagine that when it comes to reducing THD, the cure could be worse than the ailment in some cases...

[dave walsh]

The presenter [in the studio] pronounced Lichen as "Litchen". I always thought it was "Liken". A small point perhaps I didn't find the sound quality too distracting. Getting to hear it at all trumps the reproduction for me! I know the 4xtra people do process the material they find. It may just be a matter of what we are used to individually. After moving on from the early days when a 9k top response was the "norm" I've never been too concerned but as you say, there may be room for improvement. To be honest I'm often slightly irritated when there are profuse apologies for the quality, eg some Paul Temple episode when I haven't thought it to be that bad. Sometimes it might even add to the period atmospheric. This is a channel dedicated to Vintage Material after all. Lets hope they don't try to wipe it again

Dave W

[Ted Kendall]

There was very little objectionable about BBC tape recordings from the early 1950s onward. Indeed, nitrate disc recordings often sounded excellent as well. Negligible wow and flutter, a frequency range up to 15kHz and s/n ratio around 60dB were the norm.

The trouble starts when said programmes are recorded off-air. At its best, off-air recording is practically indistinguishable from source. For example, I have a tape of an opera broadcast from Glyndebourne in 1956 which might as well have come straight from continuity output. Granted, this is at 7.5 ips, recorded on a machine which accommodated NAB reels and fed from an FM receiver, but it shows what could be achieved. More common is AM reception, with all the deficiencies that involves, recorded at slow speed on a domestic machine, which could be anything from a Ferrograph or Grundig to something like a Walter(!).

I butted up against these problems when mastering The Goon Show and other series for the BBC CD releases.The most acute case is where a cut line has to be inserted in a studio tape from a domestic recording and an obvious change of quality across the join has to be avoided. CEDAR hiss reduction is usually required, along with careful equalisation to match the tonal character of the two recordings. Speed has to match within close limits, too. Distortion can also be disturbing but is difficult to deal with, despite the claims of software manufacturers. Straightforward hard clipping is relatively easy, and CEDAR have an effective tool for this, but the softer overloads and intermodulation often associated with domestic tape are to all intents and purposes baked in.

In sound restoration, it is useful to bear in mind that all the techniques available can push a given piece of audio so far up the slope of quality - the lower it starts, the lower it will finish, and the over-processing so popular in some quarters is more distracting than the original faults.

I would dispute that there is anything wrong with the original sound quality of Round The Horne, or indeed any of BBC comedy output of the time. Granted, there are rough edges and odd technical funnies, but is isn't fair to criticise it on the basis of domestic recordings. It is easy to berate the BBC for not keeping your favourite programme in broadcast quality, but the plain truth is that this just wasn't practical with the technology of the time - it was estimated in the 1970s that to keep everything would have required a storage facility the size of W1.

I'm not involved in Hidden Treasures, but I'd bet that the budgets available for restoration aren't very large, and the major difference between a so-so job and a good one is time and skill rather than software. The latter is a tool, not a panacea, and we all know what happens when we try to bang nails in with a screwdriver...

I've now had a quick listen to the programme in question on Sounds. I'd say this is probably good AM reception with treble filtering, or FM indifferently recorded at slow speed. Either way, there's boxiness aplenty and heavy processing for hiss. It warrants the apology, in my view.

[Ted Kendall]

Originally Posted by dave walsh

I always thought it was "Liken".

Agreed.

[Pfraser]

Yes I thought it was 'liken'!

Having spent many hours painstakingly getting audio files 'just right', I tend to agree that editing suite time is likely a key factor at 4Extra.

My own focus in this field has involved source material and technology from the 1980s to now. I am therefore probably more sensitive to 'rougher' recordings. I did wonder if the programme would sound more agreeable if heard on a 1950s MW set. I used a modern DAB portable.

Most interesting info on DAT etc at the BBC. Thanks. Former colleagues have expressed a preference for MiniDisc as an on-air source, finding DAT difficult to reliably cue.

[Brigham]

It was 'litchen' when I first heard it, as a kid. Colonel Breen pronounces it as such in Quatermass.

This was before 'poncy' foreign pronunciations became fashionable. I suspect that 'li-ken' comes via the USA.

Have to agree about transcription technology. Whatever the standards of the time, the BBC was at the top end.

[lesmw0sec]

One of my hates is the over application of noise gates on old film sound, which results in some quiet speech getting totally lost. I'd rather hear the words and put up with the hiss & bubble!

[Ted Kendall]

Broadband expansion/gating is a blunt instrument at best. It affects ambience as well as noise, leading to the impression of the sound being disembodied - as somebody memorably put it, "there's no there there!". More sophisticated noise reduction software allows the frequencies where noise is most objectionable to be treated, leaving ambience intact.

[lesmw0sec]

Originally Posted by Ted Kendall

Broadband expansion/gating is a blunt instrument at best. It affects ambience as well as noise, leading to the impression of the sound being disembodied - as somebody memorably put it, "there's no there there!". More sophisticated noise reduction software allows the frequencies where noise is most objectionable to be treated, leaving ambience intact.

That assumes that someone might find it objectionable. I just wish they would leave it alone. If I want such modification, I'd prefer to do it myself!

[jamesperrett]

Originally Posted by lesmw0sec

That assumes that someone might find it objectionable. I just wish they would leave it alone. If I want such modification, I'd prefer to do it myself!

With modern tools, if noise reduction is applied properly you shouldn't be aware that it has been used (or at least that's the approach that I take).

[Ted Kendall]

Originally Posted by jamesperrett

With modern tools, if noise reduction is applied properly you shouldn't be aware that it has been used (or at least that's the approach that I take).

Exactly. The beauty of frequency-selective noise reduction is that it can be used to make the noise spectrum more white or pink in character and therefore much less intrusive than the raw reduction figure would suggest.

[Graham G3ZVT]

I'm sure I can hear similar artefacts on the broadcast similar to my own ham-fisted attempts at noise reduction using Audacity or Cool Edit Pro.
So maybe the raw audio sounded very much worse.

Of course with modern AI you can always recover the original sound. (he said with tongue firmly in cheek.)