Status
10 entries
Results
6Finished: the setup, the numbers and what they mean.
-
NL-2610.006Result[nl.SA]
Using Whisper to check text-to-speech takes
The Nook team
For a 40-second film introducing Nook, we generated the voiceover locally with Supertonic, an open text-to-speech model, running on audio.cpp. Its output differs from run to run, and some takes stumble. Rather than listen to every take, we transcribed each one with Whisper and kept the take whose transcript matched the script exactly. The same check showed which words the voice gets wrong: "AI" and "GPU" were read correctly, "A.I." was not, and "Nook" in the middle of a sentence often came out as "nuke". The transcripts' word timings then helped time the film's visuals to the voice.
- Comments
- Supertonic 3, voice F1, on audio.cpp; Whisper small on whisper.cpp's CUDA build. The number of takes was not recorded.
- Subjects
- Speech and audio (nl.SA); Methods (nl.MT)
-
NL-2610.005Result[nl.DO]
Editing text in a PDF without damaging the rest of the page
The Nook team
Nook's PDF editor changes text where it stands, in the same font, using PDFium. In testing, a PDF made with an online PDF tool came back from an edit with its icons and numbers turned black: when PDFium rewrites a page, it can drop colours defined by ICC colour profiles, and gradients. Nook now draws the page before and after every edit and compares the pixels outside the edited words. If anything else changed, the edit is taken back and tried a safer way, and if no way is safe the page is left as it was. For words inside a scan, Nook reads them with Windows' built-in OCR and matches the lettering against 40 installed typefaces, or reuses the letters already on the page.
- Comments
- PDFium chromium/7881 through pdfium-render 0.9.4; Windows.Media.Ocr. Checked by two live tests.
- Subjects
- Documents (nl.DO)
-
NL-2610.004Result[nl.MM]
An "8 GB" graphics card reports 7.99 GiB, and the desktop may be using half of it
The Nook team
When we added more video models to Nook, each listed with the video memory (VRAM) it needs, none of the 8 GB models was marked as fitting our 8 GB card. The RTX 4060 reports 8188 MiB, which is 7.99 GiB, so a strict comparison failed by a few megabytes. Nook now allows half a gigabyte of slack when it compares a model's needs with the card. While testing, we also found about 4.7 GB of the card already in use by desktop apps before any model loaded. The number on the box says little about what a model will actually find free.
- Comments
- RTX 4060 8 GB, Windows 11. Two readings from one machine, not a survey.
- Subjects
- Memory (nl.MM); Images and video (nl.VI)
-
NL-2610.003Result[nl.SM]
A 132 MB embedding model picks the right tool from a plain request
The Nook team
In Nook you ask for a Nooklet, a small ready-made job such as transcribing a recording or editing a PDF, in plain words, and a model picks it. We wanted that to work on any computer with nothing to set up, so we used multilingual-e5-small, an embedding model of 132 MB in 8 bits, running on the processor through llama.cpp. It compares the request with example requests written for each Nooklet. It passed all 30 cases of our live test, including a request in Turkish, and requests that matched no Nooklet scored 0.83 to 0.85 against the 0.88 we treat as a sure match. The margin is narrow, which is why near-misses ask before they act.
- Comments
- Processor only, llama.cpp CPU build. 30 test cases written by us; thresholds set by hand.
- Subjects
- Small models (nl.SM)
-
NL-2610.002Result[nl.MM]
On Windows, a model that does not fit in video memory does not fail. It crawls.
The Nook team
A translation in Nook sat at "Speaking · 0 of 1" for minutes, with no error. Whisper, the speech-to-text model, was still loaded when the voice model started, and together they did not fit in the card's 8 GB of video memory (VRAM). Instead of failing, the job spilled into the computer's ordinary memory and slowed to a crawl. Windows does this for programs that use Vulkan, which our voice engine did at the time, and NVIDIA's driver has done it for CUDA programs since version 536.40. Since Nook 0.5.5 the app unloads idle models before the voice speaks. This entry shows how to recognise a spill and, for CUDA, how to turn it back into an error.
- Comments
- RTX 4060 8 GB, Windows 11. Seen in normal use; not timed.
- Subjects
- Memory (nl.MM); Performance (nl.PF)
-
NL-2610.001Result[nl.EB]
audio.cpp on an RTX 4060: the CUDA build speaks more than twice as fast as Vulkan
The Nook team
Nook's speech translator can speak a translation in the speaker's own voice, using the audio.cpp engine. We first shipped audio.cpp's Vulkan build, which runs on any graphics card, then built the CUDA version for NVIDIA cards and timed the two back to back on an RTX 4060 with 8 GB. Cloning a voice to speak three lines took 8.2 seconds with CUDA and 18.3 seconds with Vulkan, and peak video memory was about half a gigabyte lower. It is one run of each, so we treat it as a first look, but the gap was large enough to move every NVIDIA card to the CUDA build in Nook 0.5.6, at the cost of a download of about a gigabyte.
- Comments
- RTX 4060 8 GB, Windows 11. Qwen3-TTS 0.6B voice cloning. One run per build, back to back.
- Subjects
- Engines and builds (nl.EB); Speech and audio (nl.SA); Performance (nl.PF)
Ongoing
4What we are testing now, with the plan and how far it has got.
-
NL-2610.010Ongoing · Planned[nl.EB]
CUDA, Vulkan or the processor, measured properly
The Nook team
Our first look at audio.cpp (NL-2610.001) found its CUDA build more than twice as fast as its Vulkan build, from one run of each. Nook installs different builds of its engines depending on the computer: CUDA for NVIDIA cards, Vulkan for other cards, and processor builds when there is no usable card. We will time a fixed job on each build of llama.cpp, whisper.cpp, stable-diffusion.cpp and audio.cpp, many times over, and report medians, spread and peak memory, so the choice of build rests on more than one run.
- Comments
- RTX 4060 8 GB for CUDA and Vulkan, and the processor; AMD and Intel results welcome from readers.
- Subjects
- Engines and builds (nl.EB); Performance (nl.PF)
-
NL-2610.009Ongoing · Planned[nl.SA]
Whisper, size by size: speed against accuracy on real recordings
The Nook team
Nook's Transcribe and Translate speech Nooklets use Whisper, which comes in sizes from tiny to large. Bigger is more accurate and slower, but by how much, on the hardware people own? We will transcribe one set of real recordings with every size, on an RTX 4060 and on the processor alone, and report speed against word error rate, so the choice of size can rest on numbers.
- Comments
- RTX 4060 8 GB and the processor; whisper.cpp, CUDA and CPU builds.
- Subjects
- Speech and audio (nl.SA); Performance (nl.PF)
-
NL-2610.008Ongoing · Planned[nl.PF]
What does quantisation cost a coding model?
The Nook team
Quantised files are how a capable model fits on a small card: the same model, stored with fewer bits per weight. Every bit removed can cost accuracy, and download pages rarely say how much. We will run one coding model at four quantisation levels on an RTX 4060 with llama.cpp and measure file size, speed, memory, and the share of a fixed set of coding tasks each file completes, with tests deciding pass or fail.
- Comments
- RTX 4060 8 GB; llama.cpp, CUDA build.
- Subjects
- Performance (nl.PF); Memory (nl.MM)
-
NL-2610.007Ongoing · Set up[nl.VI]
Which Wan video models run on an 8 GB card?
The Nook team
Nook's Video page runs Wan models through stable-diffusion.cpp, and now offers four text-to-video models: Wan 2.1 1.3B in 8 bits, and Wan 2.2 TI2V 5B, Wan 2.1 14B and Wan 2.2 A14B in 4 bits. Only Wan 2.1 1.3B has run on our RTX 4060 so far, at about five minutes for a two-second clip. The 5B model is listed as needing 8 GB, the 14B models 16 GB. We want to know what actually runs on 8 GB, how long a clip takes, how much video memory (VRAM) each model really uses, and where each one starts to spill out of it.
- Comments
- Only the 1.3B model timed so far. RTX 4060 8 GB; stable-diffusion.cpp master-841-6b3edaa.
- Subjects
- Images and video (nl.VI); Memory (nl.MM)
Nothing matches.
Run it on your own GPU.
Create AI videos and images on your own computer with Nook. Open models, local generation, no generation credits.