Artist: Imael Angel, Cali Bee
Song: Bad Times
Album: Single
Time: 2:46
Genre: EDM / deep house (AI-originated, human re-recording)
BPM: 108
Category: warm-up, accelerations
Available: Spotify, Apple Music, Amazon Music
If you’re on social media, whether TikTok, Instagram, or Facebook, you’ve no doubt heard this song on reels and videos. I’ve been hearing it seemingly everywhere this past fall, often as the background to fitness reels, and it’s intrigued me from the very first note. Dang, what a great song for my cycling class! That bluesy voice!
But I also had a funny feeling about it. Something about it seemed a little too good to be true. Was it entirely…honest?
I looked closer and discovered that the song originated as an AI-generated production. At first, I was crushed. A part of me had previously sworn I wouldn’t play AI-generated music in my classes.
Then I read further and discovered there’s a real person behind the project. Imael Angel is the artist name of Bernadett Gángó, a Hungarian-born woman living in England. She’s a Reiki practitioner who writes her own lyrics and uses AI tools to turn them into finished songs. So there’s human creativity involved, even if the process of turning those lyrics into a complete musical production is largely handled by artificial intelligence.
And here’s another twist! The original AI-generated version of “Bad Times” took off on social media, eventually leading to a record deal with Intercord, part of Universal Music. The song was subsequently released in a new version featuring French singer Cali Bee on vocals, with human producers and musicians involved in the recording. Fréquence 3 has a more detailed account of the whole story, including how the song evolved from its AI-generated beginnings. (It’s in French, but you can always copy and paste it into a translator if you’re curious.) And here’s the best part: I found a video on Cali Bee’s own TikTok page of her actually singing the song. So that bluesy voice we’re enjoying? There’s a real person behind it after all!
I couldn’t find much background information on Cali Bee, but she is officially credited as the vocalist. That’s one mystery at least partially solved.
So, let’s talk about AI music.
I know some instructors will have strong reservations about playing it, and I understand why. There are legitimate questions about what AI-generated music means for human musicians, singers, songwriters, and producers. Will it replace paid creative work? What happens when AI systems are trained on existing music without the artists’ permission or compensation? Does it matter to us whether the person behind a song actually sang it, played the instruments, or made the musical decisions? And shouldn’t listeners have the right to know how the music they’re hearing was created?
I think those are all worthwhile questions.
Although I’ll always prefer and appreciate a real person singing and playing music, I also think it’s worth considering how much electronic music has already changed the way music is made. A lot of EDM is built from synthesized sounds, programmed beats, loops, and samples. Nobody necessarily has to be playing a traditional instrument, but there’s still a human being at the equipment making creative decisions, shaping the sound, arranging the music, and monitoring all those variables.
AI takes that a step further. It can generate sounds, vocals, arrangements, and entire musical performances from human prompts and input. That raises some very different questions about authorship, creativity, and the value of human musicianship.
I haven’t suddenly become an AI music evangelist, and my reservations haven’t disappeared. But I also don’t think I need to resolve every question about the future of music before deciding whether a particular track belongs in a cycling class.
In this case, I’m making an exception. I love the sound, I love the bluesy quality of the voice, and I can already hear how well it will work with the right intervals. Sometimes you have to separate your feelings about the technology from your appreciation of what you’re hearing.
And honestly, the lyrics aren’t always easy to make out. But that’s hardly an AI-only problem. I’m looking at you, Eddie Vedder!
So, if you can get past the song’s AI origins, I think you’ll have some fun with this one. I’ve actually got four versions for you, each with a different application.
1. Bad Times, Imael Angel & Cali Bee, 2:46, 108 bpm
The first two versions are at 108 bpm, which I find to be one of the more challenging tempos to ride to. At 108 rpm, the cadence is too fast for many riders, while the sluggish 54 rpm (half-beat) may require more resistance than some are comfortable using.
My suggestion? Accelerate the legs toward that beat rather than trying to pedal directly on it.
This would make a good second song of a warm-up for leg openers (accelerations), waking up the legs and preparing for the work to come.
Start at a cadence in the 80s or 90s rpm in Z1 or Z2. Surge the legs toward 100 rpm or more, chasing that 108 rpm beat during these intervals:
- 0:23–1:16 (53s)
- 1:36–2:43 (1m 7s)
2. Bad Times (Long Version), Imael Angel & Cali Bee, 6:40, 108 bpm
Use this longer version in a similar way to the shorter version, but later in your ride. This time, though, the accelerations aren’t simply there to prepare the legs. They’re designed to push riders toward their limits!
Start in Zone 2 or low Zone 3 (below moderate), at a cadence in the 90s rpm. Surge the legs toward 108 rpm during the following intervals. These should take riders above threshold, so expect some seriously winded faces!
- 0:23–1:16 (53s)
- 1:36–2:43 (1m 7s)
- 3:39–4:14 (35s)
- 4:50–5:17 (27s)
- 5:52–6:32 (40s)
3. Bad Times (Dance Remix), Imael Angel, 3:08, 124 bpm
Now we’re talking! This is a much easier beat for everyone to follow, and it offers a great opportunity for a progressive climb.
Start with a moderate climb in Zone 3.
- 0:15 Add R to high Z3.
- 1:17 Add R to low Z4 (below threshold).
- 1:48–2:27 (39s) Add R and stand, working at threshold or above. Then, sit back down and ease up just a bit.
- 2:42–3:06 (24s) For the final push, either add R and stand, or stay seated and surge the legs by about 5 rpm.
4. Bad Times (Emotional Deep House), Dax Deep, 3:57, 122 bpm
This is actually my preferred version, and it’s quite different from the preceding tracks. I like to use it as a ramp climb at 61 rpm.
- Start in Z3 (moderate).
- 1:06 Add R to low Z4.
- 1:54 Add R to high Z4 (threshold).
- 3:12–3:44 (32s) Add R and stand, working hard in Z5. Then ease up to the end.
You can find even more versions on Spotify, some longer, some shorter, some faster, and some slower. There are plenty of possibilities for incorporating this song into your rides.
Now I’m curious about your take on this.
Would you play a song knowing it was originally created using AI? Does it make a difference if a real person wrote the lyrics, or if the finished recording features a human singer and musicians? Where do you draw the line?
Let me know in the comments. I’d really like to hear what other instructors think!

