Teaching Qwen3-TTS to speak Swahili
Qwen3-TTS knows ten languages and Swahili is not one of them. This is how a small adapter taught it, what the tokenizer had to be told about sounds like mb and ng', and why …
Tag archive
#Text-to-Speech
Notes connected by the Text-to-Speech tag.
8
entries
→
The architecture of Qwen3-TTS-12Hz: two summed streams, 16 codes per 80 millisecond frame, 2.10 B parameters
A read of the Qwen3-TTS-12Hz checkpoint from its safetensors headers rather than its model card, covering the talker, the 15 step depth predictor, and the RVQ decoder that …
Taking the 2B Qwen3-TTS from 0.876 RTF to 0.253 RTF on Apple Silicon M4
A day of profiling a two-billion-parameter speech model on an M4 MacBook Pro, told as six wrong predictions. The largest win was nine lines of sorting, and one of my two …
Tuning CosyVoice3 in Rust: Where the Speed Actually Was
The same speech model made fast a second time, in Rust on Metal, from RTF 2.96 to 0.697. Scheduling beat kernels, and two of the measurements I steered by were wrong.
Narrating Book-Length Text on One Machine
A pipeline for turning a book-length document into narrated audio with word-level timings on one machine: three engines behind one trait, the text rules that decide quality, …
How I Made CosyVoice3 2.2× Faster on Apple Silicon
How I bounded PyTorch MPS memory growth, cut CosyVoice3 latency below real time, and moved only the dispatch-bound LLM stage to MLX.
Testing Local TTS: Pronunciation, Latency, and Release Gates
A practical quality loop for local text-to-speech: representative fixtures, contextual normalization, useful latency metrics, defect labels, and release checks.
A Field Guide to Shipping Local TTS Across Platforms
How to choose, benchmark, package, and isolate a local text-to-speech stack without treating a good demo as production evidence.
No entries are filed here yet.
Browse all notes →