How to handle samples of speeches and other non-music?

so I’ve come across this Stray Kids song (Hall of Fame) which samples Neil Armstrong’s “One small step for man” speech, and it’s got me wondering how to credit such samples in general (I’ve come across several similar examples)

the three options I can think of are:

  1. samples from artist artist relationship and simply link the artist, possibly with a descriptive Credited as
  2. samples material recording relationship and create a new recording (or in the case above, maybe an existing recording) for the speech and link that
  3. both of the above which wouldn’t give much benefit in the case of a speech (likely with a single artist), but could be good for musical samples (i.e. which guitarist or vocalist was sampled from this recording)
  4. (honorable mention to the IMDb samples relationship, which I’ve used a couple times. should probably be used in addition to whichever of the above wherever it applies)

in the following post I’ll give some examples of how I’ve handled these in the past, for additional context, feel free to do the same~


I’ve added “vocals (spoken)” credits before for stuff like this. Trouble is many speeches are not in the database so can’t just link to the recording.

Also have done “contains samples by Artist”.

Example here:

This whole release is basically constructed out of samples.


the promised examples

1 Like

what I’d be proposing is that a standalone recording could be created for the speeches and whatnot that don’t exist in the database

Adding speeches can be tricky if you don’t have a certain source. See my post above. I can recognise some of the sources, but others no idea. So all I have is that list of credits.

What do you actually add? A recording would need to be “something”. But adding a YouTube link to a video may not be where that speech was lifted from. The speech was live, but which is the actual famous recording of it?

You are almost adding a “Pseudo Recording”. A statement that “something happened” but maybe we don’t have an actual link to the original.


I don’t know that the source of the sample (YouTube, VHS, vinyl, etc.) would be too terribly important in MusicBrainz terms, and more the source of the sound (TV broadcast, sound effects library, etc.)*. this is kinda what I’ve been operating under for all the examples above that I’ve used option 2 for

taking the case I opened with (Hall of Fame and Neil Armstrong), I’d likely create a new recording (or figure out which one to use from Neil’s recordings and relationships) and link the song to both that recording and to Neil Armstrong with the recording and artist samples relationships

*this is based on the Recording docs page and the argument against format shifting meaning a new recording

edit: I suppose in some cases it might end up being a catch-all recording, say if there’s multiple recordings of the same reading of the same speech or something… I think this might be better than none at all tho (but of course in cases like this, two is probably better than one, if it’s known)

One of the puzzles to me is how other music is added to MusicBrainz. Some editors want an physical “thing” to link to. A source. They get upset if you just add “things” without source.

A “speech” is almost separate to a recording. The “Speech” happened, and then someone recorded it. The speech itself is somewhere between a Work and a Recording. This is why I threw “Pseudo Recording” at it as a term. We know Armstrong said that speech. And it then appears on so many different recordings.

“Catch-all Recording” is a good term.

I support what you are suggesting, but not sure how we categorise it. The “Speech” itself is then the parent to where every recording and sample then links to.

The example I threw in above is unknown to most, but it is a track 100% made of samples. Samples of recordings are easy to link, but speeches are trickier. Would be neat to say “this is the speech that has been sampled”.

Also note that Neil Armstrong’s speech is much longer than the single sentence often quoted. So need to work out what is actually the content of the recording you are adding. Is it just what he says on the stairs? Or the whole landing attempt? Or the whole mission?

1 Like

I love examples like this, where a couple vocal samples are warped to create instruments from it. a simple example is Satellite Pictures at an Exhibition which samples vocals from All Star, but I’ve heard at least two songs made out of Jacksfilms’ voice, which had drums and whatnot all made from his voice (don’t know that I can find the source tho), and of course YTPMVs like Big Beat Mario and Ronald McDonald insanity, using samples from Hotel Mario and Japanese McDonald’s commercials respectively

Hall of Fame only samples the popular small sentence, but I’d imagine the recording I’d be creating would represent the whole speech

This is just constructed from samples - diced and spliced. Short and long samples. Very little adjustment to the source material.

Most links could then be “sample of”. Annotation can always give more background of what is most commonly used.

Or do you have multiple “recordings”. One for “whole speech” and another for “famous bit”? this is why using real recordings is easier as a start and end is more clearly defined.

I’d only see the need for one recording, see also the usage of the Amen Break and Think Break, all linked to the “whole track” recording, even tho it’s just a short drum break from the middle of the track

This is the puzzle I mean though. What is the “one recording” for the Armstrong quote? Just the two lines commonly used, or more of the day’s recording?

MLK’s “I have a dream” speech - do you define the whole speech or just the most quoted sections?

Honestly, quoting from speeches almost sounds like, unless you know exactly the real recording used, we should be using a work for the speech and then linking to that? That has of course its own set of issues - for example, I expect the writers for speeches are often not the same people as the famous speechers.


if an artist is quoting a speech (i.e. the artist reading someone else’s speech), I’d probably handle it through works alone. a similar example I know of is The Day We Killed, in which the singer quotes a passage from a book (tho it might be better to link the works together rather than a partial recording of the book work, since it’s in the middle of the song, idk)

however, I feel this approach doesn’t make much sense for samples, tho I wouldn’t be opposed to adding and linking works as well for these cases.

(off-topic, but I’ve even considered adding work relationships for regular samples and mashups, but that might be a question for another time)


I agree with @UltimateRiff am not sure if linking direct to works make sense. My example above is literally all samples. Sample of music, samples of speeches. It would make the Recording to Work relationship unreadable as there would be so many links you could not see the actual Work for the Recording.

A list of “samples of Recording” seems to make more sense. If you needed to know which Works used by the samples then you look at the linked recording and pluck the Work relationship out from that.

(@UltimateRiff OT comment - see above paragraph. If you need the Work of a sample you’d follow the Recording and extract it from there… The artist isn’t performing the work)

My point was more that having “samples from work” combined with “samples from artist” would be more specific than just the latter when you cannot know what recording they used or adding a recording is complicated because it’s hard to even find a proper source for it. But yeah, nothing is a great fit IMO.

1 Like

proper source meaning what? the exact recording used? (i.e. a specific vinyl rip or YouTube reupload or something)

I think for the cases where we do have that information would be very few if any, and I think it would be worth having a solution for the vast majority of cases

Yes, the exact recording used, because otherwise saying it samples whatever recording gets added is just guessing and not real data.

I’m a bit confused about the references to “exact recording” here.

It should be pretty clear in most cases - for example, the moon landing has 1 very obvious “recording”, which is the one that was broadcast into millions of homes via a single television broadcast. Now whether a sample was recently taken from a youtube video or a vinyl rip where that recording was re-distributed, I don’t see how that matters. Unless an audio engineer got involved and made a new mix, it’s all the same recording, right?

Maybe I’m missing something…


Well, I’m often unsure how you’d even get the info about those recordings in order to add them, because in many cases they are actually split, cut and edited in all sort of ways. If we have the entire original speech somewhere, then sure, we can at least add that - maybe they actually took the sample from a news broadcast which only showed a little bit of the speech, but maybe we don’t care.

I don’t really edit samples so I feel out of my depth here, but I guess at the end of the day I’m not seeing the value added to the database in requiring something like the original moon landing recording/broadcast to be linked to something like a YouTube video - which presumably could be linked to the news broadcast it ripped, which is linked to the tape the news room edited from, which is linked to the original moon landing recording/broadcast anyway? All for the exact same audio being redistributed throughout the chain? :thinking:

Personally, unless the speech was specially mixed for the news broadcast, I don’t care. Or, more accurately, I would prefer it be linked to the ‘original’ recording.

1 Like