How conversation with AI on rhumba went off track

This story has significance for readers across Kenya and beyond.

It started as idle curiosity that led me to the world of mysterious Artificial Intelligence (AI) algorithms making music instead of the artistes that I, of the Baby Boomers generation, grew up with.

I have always been fan of rhumba. Today, I am part of the crazed oldies of the modern rhumba music fraternity.

I occasionally do a shuffle alone, if the occasion demands, just to spite millennials and Gen Zs. Still, I deliver serious work with a rhumba ambience in the background playing from my laptop.

My rest hours are spent surfing the internet for transcriptions and translations of rhumba music in Kiswahili, Lingala and French. I seek to avoid dancing hilariously to a funeral dirge just because I do not understand the lyrics.

Now I have an assorted functional vocabulary of Lingala and French to guard me from committing blunders at gatherings.

The prompter has always been my assistant Sammy Kaiya, an Anglican-named Muslim.

While he drives, I connect the phone to the car Bluetooth and wham! we are in Kinshasha among Lingala singing stars. The first time he slyly enquired; “These (rhumba) songs, do you understand them?”

“What?” I parried, hoping to derail him. “The songs, do you know them?”, he persisted.

Since then, I practice by explaining the meaning of the songs we listen to, henceimproving my Lingala.

Modern rumba music intrigues me with its beats — unmistakable similar guitar and vocal arrangements.

What vary are lyrics.

And I often sense fragrant violation of copyright, so I ask AI for help. What follows is stranger than fiction; the AI admits it generates this music, but it is unable to find the actual persons in violation, at least for a while.

But first things first: there are exceptions. There is no doubt that though using the prevailing guitar arrangements, Koffi Olomide is the artist behind the Lingala, Kiswahili and French hit version of “Perfum ya Bolingo”.

He is also, with his partner Cindy Le Coeur, behind “Nakotika Yo Te” and “Malembe na yo” among others.

Fally Ipupa’s “Tombe” is as authentic blend, but with similar guitar arrangements as Diamond Platinumz’s Bado.

What is intriguing is the origin under the versions of timeless sounds of East African Kiswahili music on YouTube.

“Afro” was sung and recorded by the iconic Tanzanian-Kenyan Swahili rhumba band Les Wanyika (led by founding members like John Ngereza and Sijali Salum Zuwa). “Malaika” was originally written by Tanzanian songwriter Adam Salim and first recorded and made famous by the legendary Kenyan guitarist and singer Fadhili William in the mid-1940s/1950s (later covered globally by Miriam Makeba and others).

Then there is “Susana”, a beloved classic East African guitar-band tune performed by the legendary Swahili band Simba Wanyika (founded by the Kinyonga brothers, Wilson and Peter).

When I asked AI about the artistes behind the versions of these classics, a curios merry-go-round ensured until I was able to wriggle truth. At first, there was evasion. I was provided with the songs’ transcriptions and generous translations of Lingala terms and phrases woven throughout these classic Swahili rhumba love songs.

I pressed on.

“Provide the chords or guitar tuning style of the lyrics”, I requested as if talking to a reluctant executive secretary known for rigid abstinence in the office.

“You can find the full, unabridged lyrics to all the songs from the rumba video mix, including ‘Afro’, ‘Malaika’, and the final track ‘Kipenzi Changu’, by scrolling up to the previous message”, came the answer directing me to YouTube and TikTok accounts! Unfazed, I had to change tact for a softer landing before spooking AI.

“Provide the full lyrics for the song ‘Moyo Wangu Wazimu’.”

The same stoke answer came back:“You can find the full lyrics for the emotional Swahili and Lingala love song ‘Moyo Wangu Wazimu’, which is featured in the provided rumba mix, through the following link: YouTube.”

“I need written lyrics, both in Kiswahili and Lingala”, I shot back. I was testing waters with “Moyo Wangu Wazimu” hit before finding another inroad.

After a long searching wait that felt eternity, I got my transcripts of Moyo Wangu Wazimu, except for my question “Who is the Artist?”

The chickens were coming home to roost. If AI has learned to shoot without missing, I have mastered flying without patching. “This specific mix consists of royalty-free Rumba library tracks and AI-generated Rumba concepts produced directly by African ambient channels to replicate the classic Congolese and East African Rumba style”, the AI begun the lesson.

It added: “Because it is an independent digital project rather than a commercial release by a traditional band, it does not have a single designated superstar artist (like Franco, Madilu System, or Fally Ipupa).

Instead, the instrumentation and vocals are created to emulate the iconic “Rumba vibe.” I had hit pay day by circumlocution.

“There is always a way around a difficult wall to climb — you can slice through or bring it down”, I remembered the wisdom I earned in Itumbi — the traditional healing hut for newly circumcised initiates — way back in 1968. I hilariously condoled my enthusiastic self before I was woken out of the deserved reverie by another AI’s unexpected diversion:

“If you would like”, AI implored like the grand school matron, “to explore the real-world of iconic musicians who pioneered this exact style of Swahili and Lingala music, I can recommend foundational artists such as Les Mangelepa, TPOK Jazz, or Sauti Sol (who popularised the track Suzanna).”

I wasn’t about to be side-tracked though: “Still, who are the artist(s) who generated this song?” I persisted. I wanted to understand what “AI-generated Rumba concepts produced directly by African ambient channels” meant.

The writer is a communications consultant Next week: The law, Generative AI music tools and the shadowy digital music producer

Reporting originally appeared via Nation Africa. Read the full source for additional context.