Em'

Software Engineer & Data Scientist

[Em' @whispem_io]

SF has definitely my heart 🤍

[Em' @whispem_io]

@lina238_ @Promethee_io Thanks!!

[Em' @whispem_io]

Ranked #1 on @Promethee_io 👀

[Em' @whispem_io]

@Nlacombe_

[Em' @whispem_io]

@Nlacombe_ @Promethee_io Je veux bien, merci beaucoup 🙏🏼

[Em' @whispem_io]

@ISMVV0 @Promethee_io Thanks! :)

[Em' @whispem_io]

@Teo_Goulois @Promethee_io Thank you!

[Em' @whispem_io]

Been writing a chiptune synth entirely in x86-64 assembly (NASM, Linux) — no libc, no audio lib, no FPU even.

Everything fixed-point, samples computed by hand, WAV written straight to stdout.

So far: 4 oscillators + LFSR noise, polyphony, ADSR, FM synthesis with a hand-built sine LUT, and a growing pile of effects (filters, delay, ring mod, chorus, Schroeder reverb).

Feedback on the asm very welcome — self-taught and always looking to tighten it.

Repo: https://t.co/x1SAsDOhj3

#assembly #x86_64 #chiptune #DSP #lowlevel

[Em' @whispem_io]

@Alex_Kaasten @Promethee_io Hello!
Moi c’est whispem

[Em' @whispem_io]

@Promethee_io Thank you 🔥

[Em' @whispem_io]

@Nlacombe_ @Promethee_io Thanks!

[Em' @whispem_io]

@ISMVV0 @grok @X @boardyai Let's connect!

[Em' @whispem_io]

@atwnxp @Promethee_io Je vais t’ajouter également!

Je bosse sur un synthétiseur chiptune, écrit entièrement en assembleur x86-64 :)
Si tu veux aller jeter un œil au repo voici le lien: https://t.co/YA9jCTfA4s

Et toi?

[Em' @whispem_io]

@atwnxp @Promethee_io whispem :)

[Em' @whispem_io]

@Nlacombe_ @Promethee_io Thanks!

[Em' @whispem_io]

Been using @Promethee_io for a week now, and as someone with autism + ADHD, it's one of the rare focus tools that actually helps me anchor into my sessions.

Watching my consistency fill up as a heatmap is oddly satisfying 🔥

[Em' @whispem_io]

@YounArcy @Promethee_io Thanks!

[Rafaël Amselem @Rafael_Eyth]

Le fait qu'il faille supprimer les " — " à cause de l'IA est un appauvrissement considérable de la langue française.......

[Em' @whispem_io]

sussurro — offline neural translation and voice in pure C++ with @ggerganov's ggml

→ English ⇄ Spanish ⇄ French ⇄ Italian — all 12 directions, no pivot
→ Translate + transcribe (whisper) + speak (Piper): type or speak in, read or hear out
→ 100% on-device — no server, no API, nothing leaves your machine
→ f16 / q8 / q4 quantization, Apple Silicon (Metal)
→ OPUS-MT / Marian seq2seq rebuilt from scratch on ggml (not an LLM — an actual translation model)

Short sentences translate in ~25 ms on an M5 Max.
Open source, MIT.

🔗 https://t.co/xKA5By1MeS

#cpp #ggml