gnuradio VS MeiliSearch

In gr-dtv transmitter examples for Gnuradio, I see some times people use a resampler block before the RF hardware sink. Say our sampling rate is ~9.14Msps which satisfies the Nyquist criterion because our samples are complex numbers.

2.1. Thanks for that tip, I forgot that I was able to check the source code of the WBFM Receive block. As you have said, there are mostly the same. There are some differences between how values are picked. The WBFM Receive block would be a synonym of Quadrature demod => Fir Filter (decimation => Low pass filter) => FM Deemphasis. 2.3. My question there is why 10 and not 20 or 100. I understand that the idea is to reduce the sample rate asap, but what I don't understand is why those values were picked and how can I understand what would be the "correct" or "best" value. 2.4. I'm not fully understanding what you said. If I check the WB FM recieve source code the values that are supplied as the cutoff freq and transition width of the Low pass filter differ from the one of the example. The webfm would apply a sample rate / decimation / 2 - sample rate / decimation / 32 as a cutoff freq and a sample rate / decimation / 32 as a transition transition width. Calculating those values would end up in different that the ones supplied in this second example. Again, is there a rule of thumb to pick these values?.

Gaussian filter is used only on the tx side, so specifying bt in the receiver makes no sense. Take a look at gfsk mod/demod blocks implementation: https://github.com/gnuradio/gnuradio/blob/master/gr-digital/python/digital/gfsk.py

GNU Radio

Here's the filter coefficients used for the GNU-Radio interpolator block to get you started. This is a 7th order interpolator (i.e., 8 FIR taps) with very good performance. Each "row" of the array sets the delay in steps of sample_time / 128.

From their README: “open-source software development toolkit that provides signal processing blocks to implement software radios.” https://github.com/gnuradio/gnuradio

We needed to introduce a new service for search. As we settled on using meilisearch, we needed a way to sync updates on our models with the records in meilisearch. We could've continued to use callbacks but we needed something better.

Meilisearch

In this case, a good strategy is to use vector search only when the keyword/prefix search returns none or just a small number of results. A good candidate for this is MeiliSearch. It uses custom ranking rules to provide results as fast as the user can type.

With the VertexAI endpoint set up and tested, our next step is to work with LibreChat. LibreChat is an open-source ChatGPT clone that can integrate with various AI models, including the PaLM 2 models via the VertexAI API. It's built using React, MongoDB, and Meilisearch technologies.

Meilisearch seems like it is the best open source option.

https://www.meilisearch.com/

Starting with v1.3, you can use Meilisearch as a vector store. Meilisearch allows you to store vector embeddings alongside your documents conveniently. You will need to create the vector embeddings using your third-party tool of choice (Hugging Face, OpenAI). As we published the first v1.3 release candidate, you can try out vector search today.

I work at Meilisearch, an open-source search engine built in Rust. 🦀

https://www.meilisearch.com/ https://github.com/meilisearch