-
Notifications
You must be signed in to change notification settings - Fork 602
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add Speech to Speech example notebook, closes #789
- Loading branch information
1 parent
f3138c2
commit 67321b0
Showing
4 changed files
with
265 additions
and
2 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,262 @@ | ||
| { | ||
| "cells": [ | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# Speech to Speech RAG\n", | ||
| "\n", | ||
| "[txtai](https://github.com/neuml/txtai) is an all-in-one embeddings database for semantic search, LLM orchestration and language model workflows.\n", | ||
| "\n", | ||
| "There are many articles, notebooks and examples covering how to perform vector search and/or retrieval augmented generation (RAG) with txtai. A lesser known component of txtai is it's built-in workflow component.\n", | ||
| "\n", | ||
| "Workflows are a simple yet powerful construct that takes a callable and returns elements. Workflows enable efficient processing of pipeline data. Workflows are streaming by nature and work on data in batches. This allows large volumes of data to be processed efficiently.\n", | ||
| "\n", | ||
| "This notebook will demonstrate how to to build a Speech to Speech (S2S) workflow with txtai.\n", | ||
| "\n", | ||
| "_Note: This process is intended to run on local machines due to it's use of input and output audio devices._" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# Install dependencies\n", | ||
| "\n", | ||
| "Install `txtai` and all dependencies." | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", | ||
| "execution_count": null, | ||
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "%%capture\n", | ||
| "!pip install git+https://github.com/neuml/txtai#egg=txtai[pipeline-audio] autoawq" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# Define the S2S RAG Workflow\n", | ||
| "\n", | ||
| "The next section defines the Speech to Speech (S2S) RAG workflow. The objective of this workflow is to respond to a user request in near real-time.\n", | ||
| "\n", | ||
| "txtai supports workflow definitions in Python and with YAML. We'll cover both methods.\n", | ||
| "\n", | ||
| "The S2S workflow below starts with a microphone pipeline, which streams and processes input audio. The microphone pipeline has voice activity detection (VAD) built-in. When speech is detected, the pipeline returns the captured audio data. Next, the speech is transcribed to text and then passed to a RAG pipeline prompt. Finally, the RAG result is run through a text to speech (TTS) pipeline and streamed to an output audio device." | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", | ||
| "execution_count": null, | ||
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "import logging\n", | ||
| "\n", | ||
| "from txtai import Embeddings, RAG\n", | ||
| "from txtai.pipeline import AudioStream, Microphone, TextToSpeech, Transcription\n", | ||
| "from txtai.workflow import Workflow, StreamTask, Task\n", | ||
| "\n", | ||
| "# Enable DEBUG logging\n", | ||
| "logging.basicConfig()\n", | ||
| "logging.getLogger().setLevel(logging.DEBUG)\n", | ||
| "\n", | ||
| "# Microphone\n", | ||
| "microphone = Microphone()\n", | ||
| "\n", | ||
| "# Transcription\n", | ||
| "transcribe = Transcription(\"distil-whisper/distil-large-v3\")\n", | ||
| "\n", | ||
| "# Embeddings database\n", | ||
| "embeddings = Embeddings()\n", | ||
| "embeddings.load(provider=\"huggingface-hub\", container=\"neuml/txtai-wikipedia\")\n", | ||
| "\n", | ||
| "# Define prompt template\n", | ||
| "template = \"\"\"\n", | ||
| "Answer the following question using only the context below. Only include information\n", | ||
| "specifically discussed. Answer the question without explaining how you found the answer.\n", | ||
| "\n", | ||
| "question: {question}\n", | ||
| "context: {context}\"\"\"\n", | ||
| "\n", | ||
| "# Create RAG pipeline\n", | ||
| "rag = RAG(\n", | ||
| " embeddings,\n", | ||
| " \"hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4\",\n", | ||
| " system=\"You are a friendly assistant. You answer questions from users.\",\n", | ||
| " template=template,\n", | ||
| " context=10\n", | ||
| ")\n", | ||
| "\n", | ||
| "# Text to speech\n", | ||
| "tts = TextToSpeech(\"neuml/vctk-vits-onnx\")\n", | ||
| "\n", | ||
| "# Audio stream\n", | ||
| "audiostream = AudioStream()\n", | ||
| "\n", | ||
| "# Define speech to speech workflow\n", | ||
| "workflow = Workflow(tasks=[\n", | ||
| " Task(action=microphone),\n", | ||
| " Task(action=transcribe, unpack=False),\n", | ||
| " StreamTask(action=lambda x: rag(x, maxlength=4096, stream=True), batch=True),\n", | ||
| " StreamTask(action=lambda x: tts(x, stream=True, speaker=15), batch=True),\n", | ||
| " StreamTask(action=audiostream, batch=True)\n", | ||
| "])\n", | ||
| "\n", | ||
| "while True:\n", | ||
| " print(\"Waiting for input...\")\n", | ||
| " list(workflow([None]))" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "Given that the input and outputs are audio, you'll have to use your imagination if you're reading this as an article.\n", | ||
| "\n", | ||
| "[Check out this video](https://www.youtube.com/watch?v=tH8QWwkVMKA) to see the workflow in action! The following examples are run:\n", | ||
| "\n", | ||
| "- Tell me about the Roman Empire\n", | ||
| "- Explain how faster than light travel could work\n", | ||
| "- Write a short poem about the Vikings\n", | ||
| "- Tell me about the Roman Empire in French" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# S2S Workflow in YAML\n", | ||
| "\n", | ||
| "A crucial feature of txtai workflows is that they can be defined with YAML. This enables building workflows in a low-code and/or no-code setting. These YAML workflows can then be \"dockerized\" and run.\n", | ||
| "\n", | ||
| "Let's define the same workflow below." | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", | ||
| "execution_count": null, | ||
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "%%writefile s2s.yml\n", | ||
| "# Microphone\n", | ||
| "microphone:\n", | ||
| "\n", | ||
| "# Transcription\n", | ||
| "transcription:\n", | ||
| " path: distil-whisper/distil-large-v3\n", | ||
| "\n", | ||
| "# Embeddings database\n", | ||
| "cloud:\n", | ||
| " provider: huggingface-hub\n", | ||
| " container: neuml/txtai-wikipedia\n", | ||
| "\n", | ||
| "embeddings:\n", | ||
| "\n", | ||
| "# RAG\n", | ||
| "rag:\n", | ||
| " path: \"hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4\"\n", | ||
| " system: You are a friendly assistant. You answer questions from users.\n", | ||
| " template: |\n", | ||
| " Answer the following question using only the context below. Only include information\n", | ||
| " specifically discussed. Answer the question without explaining how you found the answer.\n", | ||
| "\n", | ||
| " question: {question}\n", | ||
| " context: {context}\n", | ||
| " context: 10\n", | ||
| "\n", | ||
| "# TTS\n", | ||
| "texttospeech:\n", | ||
| " path: neuml/vctk-vits-onnx\n", | ||
| "\n", | ||
| "# AudioStream\n", | ||
| "audiostream:\n", | ||
| "\n", | ||
| "# Speech to Speech Chat workflow\n", | ||
| "workflow:\n", | ||
| " s2s:\n", | ||
| " tasks:\n", | ||
| " - microphone\n", | ||
| " - action: transcription\n", | ||
| " unpack: False\n", | ||
| " - task: stream\n", | ||
| " action: rag\n", | ||
| " args:\n", | ||
| " maxlength: 4096\n", | ||
| " stream: True\n", | ||
| " batch: True\n", | ||
| " - task: stream\n", | ||
| " action: texttospeech\n", | ||
| " args:\n", | ||
| " stream: True\n", | ||
| " speaker: 15\n", | ||
| " batch: True\n", | ||
| " - task: stream\n", | ||
| " action: audiostream\n", | ||
| " batch: True" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", | ||
| "execution_count": null, | ||
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "from txtai import Application\n", | ||
| "\n", | ||
| "app = Application(\"s2s.yml\")\n", | ||
| "while True:\n", | ||
| " print(\"Waiting for input...\")\n", | ||
| " list(app.workflow(\"s2s\", [None]))" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "Once again, the same idea, just a different way to do it. In the video demo, the following query was asked.\n", | ||
| "\n", | ||
| "- As a Patriots fan, who would you guess is my favorite quarterback of all time is?\n", | ||
| "- I'm tall and run fast, what do you think the best soccer position for me is?\n", | ||
| "- I run slow, what do you think the best soccer position for me is?\n", | ||
| "\n", | ||
| "With YAML workflows, it's possible to fully define the process outside of code such as with a web interface. Perhaps someday we'll see this with [txtai.cloud](https://txtai.cloud) 😀" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# Wrapping up\n", | ||
| "\n", | ||
| "This notebook demonstrated how to build a Speech to Speech (S2S) workflow with txtai. While the workflow uses an off-the-shelf embeddings database, a custom embeddings database can easily be swapped in. From there, we have S2S with our own data!" | ||
| ] | ||
| } | ||
| ], | ||
| "metadata": { | ||
| "kernelspec": { | ||
| "display_name": "local", | ||
| "language": "python", | ||
| "name": "python3" | ||
| }, | ||
| "language_info": { | ||
| "codemirror_mode": { | ||
| "name": "ipython", | ||
| "version": 3 | ||
| }, | ||
| "file_extension": ".py", | ||
| "mimetype": "text/x-python", | ||
| "name": "python", | ||
| "nbconvert_exporter": "python", | ||
| "pygments_lexer": "ipython3", | ||
| "version": "3.8.20" | ||
| } | ||
| }, | ||
| "nbformat": 4, | ||
| "nbformat_minor": 2 | ||
| } |