For longer samples like speech phrases, the two boxes behave quite differently once you look past “it can play samples”.
Analog Rytm II and long samples
Key constraints and quirks:
1. Hard length limit and small RAM
• Max recording / sample length is about 33 seconds per sample. 
• Per-project sample RAM is 64 MB, mono only, with up to 127 sample slots. 
• Samples are converted to 16-bit / 48 kHz mono on import. 
For spoken bits: 33 s is fine, but you cannot do multi-minute speeches or many long stereo ambiences.
2. Each long sample consumes an analog voice
• Every track is a combined analog drum engine + sample layer; the sample always goes through that voice’s analog filter, overdrive, etc. 
• If that voice is retriggered (e.g. another trig on the same track or another track sharing that voice, depending on voice allocation), it can cut off your long speech tail.
• In practice, you almost always dedicate one track/voice purely to speech if you want longer phrases.
3. No time-stretching
• Rytm plays back samples with pitch/tune and a basic envelope only; no time-stretch / warp. 
• For speech, that means:
• Changing tempo without artifacts = awkward (you need to re-sample at the right BPM).
• Big tuning moves quickly sound like tape-speed effects, not neutral “faster/slower speech”.
4. Chopping speech is powerful but “manual”
• You can still do nice speech-cut stuff by:
• Putting the whole phrase on one track,
• Using parameter-locked START (and maybe LEN) per step to jump to different words / syllables.
• Works well, but sample management on Rytm is more cramped and mono-only. 
5. Why you might want Rytm for speech
• Long samples get the full analog voice treatment: filter, analog overdrive, analog compressor, plus send FX. 
• Spoken bits can become very characterful: saturated “radio voice”, crushed vox stabs, etc.
• It’s great when speech is just one more percussive / textural element, not the structural backbone of a track.
Digitakt II and long samples
Now the same use-case on Digitakt II (DT2):
1. Twice the per-sample time, much bigger RAM
• Max sampling time is 66 seconds per sample. 
• Per-project sample RAM is 400 MB, with about 20 GB on the +Drive. 
• Stereo sampling and playback are supported on all audio tracks. 
For speech: you can host a lot more long phrases, including stereo ambience, intros, field-recorded monologues, etc.
2. Dedicated digital voices; no analog voice stealing
• Up to 16 tracks can be audio tracks, each with its own sample. 
• A long speech on track 1 will not steal an analog voice from drums, because everything is digital and each audio track is its own voice.
• Much easier to have several speech tracks (e.g. narrator, background chatter, one-shot quotes) running alongside drums.
3. Time-stretching and playback “machines”
• DT2 has multiple SRC “machines” for sample playback (Oneshot, Repitch, Stretch, Werp, Grid etc.). 
• There is proper time-stretch / warping, significantly updated over the original Digitakt. 
• For speech this is important:
• You can keep pitch natural while nudging tempo.
• You can lock speech to the grid more gracefully, especially for voice-over, spoken word over beats, etc.
4. Better “long-phrase” sequencing ergonomics
• Patterns up to 128 steps, with per-track length and time-scale. 
• Long one-shots will happily play across patterns as long as envelopes allow; and the manual explicitly notes that pressing STOP once leaves long samples ringing, double-STOP kills them. 
• Combined with trig conditions, Euclidean generator, and parameter locks, you can:
• Trigger speech rarely while drums keep going.
• Slice a long monologue across a pattern using START/LEN locks, similar to Rytm, but with more steps and better visual feedback.
5. Sample-heavy sets and limitations
• DT2 is still RAM-based, not streaming from card; if your set is built from multiple multi-minute stems, you can run into limits, and people do point out that for “pure long-stem playback” something like Octatrack/SP-404 can be more suitable. 
• But for “bits of speech” – intros, drops, cut-up phrases – 66 s per sample and 400 MB RAM are very comfortable.
Narrow verdict: speech / long phrases on Rytm II vs Digitakt II
For the specific use of long speech bits (not just drum one-shots):
• **Analog Rytm II**
• Pros:
• Analog coloration on speech (filter, drive, compressor) is unique and very musical.
• Great for short–medium phrases that you want to mangle into the drums.
• Cons:
• 33 s per sample, 64 MB mono RAM, and 8 analog voices shared with your drums mean you quickly hit limits if you rely heavily on long samples.
• No time-stretch; changing song tempo or rearranging speech tightly to grid requires more prep work.
• **Digitakt II**
• Pros:
• 66 s per sample, 400 MB RAM, 16 audio tracks, stereo support: far better suited to hosting multiple long speech phrases in a project.
• Time-stretch / warp machines make it easier to keep speech intelligible while matching tempo.
• No analog voice sharing – long samples don’t choke your drums.
• Cons:
• No analog overdrive/filter per voice; if you love the “Rytm crust” on speech, you’d need to recreate that with DT2’s digital FX or external gear.
• Still not a full “backing-track player” like an Octatrack if you want multi-minute continuous narration.
If your main interest is having several long spoken phrases available, cut-up and re-arranged over your beats, Digitakt II is clearly the stronger tool. I’d see Analog Rytm II more as “one great, analog-processed speech track per project” used as a character element rather than as your main long-sample workstation.
