Document

I want to input a youtube video, then get a collection of music samples returned to me (.wav files in a zip). I think we need to rip the audio file from YouTube, then extract stems (instrumentals, vocals), and then get the beatpoints/tempo. Then we need to chop up samples from vocals and beatpoints. I think with this approach we can practically extract as many audio samples as we want, but we may get too many. Therefore, we may want to only focus on the key parts of the song (which are likely to give the best samples), namely the initial song buildup and the chorus. We should be able to visually "see" these on a spectrograph (the quietest and loudest parts of the song). So I think we should begin by ripping the song, then immediately running the spectrograph, then select 3 sections of the song - perhaps around 16 beats in length - one that is the initial song buildup (basically the first 16 beats of the song), then one that is the hook/chorus of the song (the loudest part), and then one that is the end of the song (the final 16 beats). Doing this separately for the vocal and instrumental stems will give us 6 total audio files to do this with. Then we want to chopup the samples using SOX, perhaps chopping based on beat length (if BPM = 90bpm, then 1 beat is 0.66secs) and beat position, and chopping maybe a mix of 1/2/4/8 beat samples. This may still generate too many samples, so perhaps there is a way to filter down, but I will let you think about that. Finally we will want to save all the files to a server and zip them up.

final-output-and-component-outputs

Thank you for your detailed task description. Let me address your questions one by one to ensure that I understand the final output and the components required to achieve it.

a) Final output:
The final output of this task would be a zip file containing a collection of .wav audio samples extracted from a specific YouTube video. These samples would be derived from the vocal and instrumental stems of the song, focusing on key parts such as the initial buildup, the chorus, and the end of the song. The samples would be chopped based on beat length and position, with a mix of 1, 2, 4, and 8 beat samples.

b) Components/outputs needed:

1. YouTube video URL: The input to the system, specifying the video from which the audio samples will be extracted.

2. Audio file (mp3 or wav): The audio extracted from the YouTube video, which will be used for further processing.

3. Vocal and instrumental stems: The separated vocal and instrumental tracks from the original audio file. These stems will be used to generate samples independently.

4. Spectrograph data: Visual representation of the audio's frequency spectrum over time, used to identify the key parts of the song (initial buildup, chorus, and end).

5. Beatpoints and tempo (BPM) data: Information about the rhythm and tempo of the song, used to determine the length and position of the samples to be chopped.

6. Selected audio sections: Three 16-beat sections from each stem (vocal and instrumental), corresponding to the initial buildup, chorus, and end of the song. These sections will be used for chopping samples.

7. Chopped audio samples: The final audio samples generated by chopping the selected audio sections based on beat length and position. These samples will be a mix of 1, 2, 4, and 8 beat lengths.

8. Filtered audio samples (optional): If the number of generated samples is too large, an additional filtering step may be applied to reduce the number of samples based on specific criteria (e.g., audio quality, uniqueness).

9. Zip file: The final deliverable, containing all the generated audio samples in .wav format, ready for download or further use.

To summarize, the core components required are the YouTube video URL, the extracted audio file, vocal and instrumental stems, spectrograph data, beatpoints and tempo data, selected audio sections, and the chopped audio samples. These components will be processed and combined to generate the final zip file containing the desired audio samples.

claude 3 - extract final output

extract component outputs

Claude - research-ideas-and-input-variables

Thank you for the detailed description of the task you want to automate using an AI Agent. Before we proceed with building the workflow, let me address your questions to ensure we have a clear understanding of the final output and the necessary inputs.

a) Questions to clarify the final output:

1. How many music samples do you expect to be generated from a single YouTube video?
2. What specific file format do you require for the music samples (e.g., .wav, .mp3, .aiff)?
3. Should the music samples be labeled or named in a specific way to indicate their origin or characteristics?
4. Do you have any preferences regarding the length of each music sample (e.g., a specific number of seconds or a range)?
5. Are there any additional criteria for filtering or selecting the most suitable music samples from the generated set?

b) Inputs required before launching the task completion process:

1. YouTube video URL: The user should provide the URL of the YouTube video they want to extract music samples from.
2. Desired number of music samples: The user should specify the total number of music samples they expect to receive from the given video.
3. Output file format: The user should confirm the desired file format for the music samples (e.g., .wav, .mp3, .aiff).
4. Sample length preferences: The user may provide guidelines on the desired length of each music sample, either as a specific duration or a range.
5. Filtering criteria (optional): If the user has any specific criteria for filtering or selecting the most suitable music samples, they should provide those guidelines.

By addressing these questions and gathering the necessary inputs, we can ensure that the AI Agent has a clear understanding of the final output and can tailor the workflow accordingly. This will help prevent any misinterpretation and ensure that the generated music samples align with your expectations.

Once we have clarified these aspects, we can proceed with building out the automated workflow for extracting, processing, and delivering the music samples from the given YouTube video.

GPT - research-ideas-and-input-variables

a) Some of the questions to clarify the final output could be:

1. What is the intended use of these samples? Are they for personal use or commercial use? This information could shape the legal considerations involved in the extraction process.
2. Do you have a specific audio quality for the output files? For instance, should they be high-quality or a standard setting would suffice?
3. Are there specific audio characteristics you are interested in? For instance, are you looking for samples with certain instrument sounds, vocals only, or beats?
4. What would be a reasonable quantity of samples generated from each section of the song (initial buildup, hook/chorus, end)? 
5. What is your preferred beat count in each sample? You mentioned 1/2/4/8. Should there be an even spread across these or should some be prioritized?
6. Is there a certain length for these samples that is optimum for you – for instance, do all samples need to be in the 0.66 sec range or could they vary according to the beat and section of the song?
7. Are there any restrictions on the volume, pitch or other acoustic parameters for the samples?
8. Additionally, is there a size limit for the final output .zip file, as this could influence the quantity and quality of the samples we generate?

b) Before launching the task completion process, the task should accept:

1. The specific YouTube video link from where the audio will be ripped.
2. A preferred format of the final audio files if there's any other than .wav.
3. The server details where the final files will be stored.
4. Any specific instructions or requirements about the sections of the song to be used (initial buildup, hook/chorus, end).
5. Definitions of what constitutes the quietest and loudest parts of a song, if different from common definitions.
6. The selection of beat counts for chopping up the samples (1/2/4/8 or any other).
7. Any specific parameters for filtering down the samples if the initial process generates too many.

template-ideas

Based on your description, here are a few templates and sources of information that could help define what a successful output for your automated music sampling workflow might look like:

1. Product descriptions and feature lists of existing sample packs or sound libraries on music production marketplaces like Splice, Loopmasters, or Beatport. These descriptions often detail the number of samples, file formats, lengths, and how the samples are organized. This can give a clear idea of industry standards and customer expectations for sample packs.

2. Screenshots or videos of the user interfaces of popular music sampling and chopping software like Ableton Live, FL Studio, or Maschine. Seeing how professional tools organize and present samples visually can provide inspiration for your own workflow's output format.

3. README files or documentation from open-source music sampling projects on GitHub. While the code itself may not be interpretable by the AI agent, the written descriptions, file structures, and any included screenshots could outline what types of outputs are produced.

4. Transcriptions of YouTube tutorials or demos by music producers showing their sampling process and the resulting sample collections they create. The narration may describe key characteristics of useful samples and how they are sourced from songs.

5. Written interviews or articles where professional producers break down their sampling techniques and the qualities they look for in a good sample. These insights can help prioritize what attributes to focus on in your automatically generated samples.

6. Image files showing waveforms or spectrographs of individual samples used in popular songs. Analyzing the visual patterns of these proven, high-quality samples could guide your audio processing steps.

7. Metadata or specifications from sample packs that include tempo and key information for each sample. This can demonstrate how to make your output more useful and organized for producers.

By aggregating information from these various template sources, the AI agent should start to form a more concrete understanding of what your music sampling workflow's output needs to include to be successful and valuable to end users. The agent can then keep these templates in mind as a reference while processing the steps to transform a YouTube video into a collection of compatible, useful audio samples.

claude 3 - searches

Here are 5 specific Google search keyphrases to find relevant templates and examples for your automated music sampling workflow:

1. "splice sample pack vocal chops bpm key wav" - This search should surface product pages for sample packs on Splice that contain chopped vocal samples labeled with tempo and key information, giving a clear template for how to structure your output.

2. "ableton live slicing vocals to midi tutorial" - Searching for Ableton tutorials on slicing vocal samples should yield videos showing the process and the resulting chopped samples in the software's interface, providing visual inspiration for your workflow.

3. "open source python code extract samples from song github" - Looking for open-source Python projects on GitHub related to extracting samples from songs should uncover repositories with helpful README documentation and code structure examples.

4. "fl studio tutorial chop samples from song youtube" - This search aims to find YouTube videos of FL Studio tutorials demonstrating how to chop samples from a song, which may include useful narration and before/after examples of the sampling process.

5. "reddit ama professional producer share tip finding best samples in song" - Searching for Reddit Ask Me Anything threads or interviews where professional music producers share their techniques for finding high-quality samples in songs could surface insightful written descriptions to guide your approach.

These targeted searches should help uncover specific examples and templates that can refine your understanding of what constitutes a successful output for your automated music sampling workflow. The results should include a mix of visual, auditory, and written references to guide the development of your solution.

subagent-summary

A) FULL TASK SUMMARY: Extract a collection of music samples in .wav format from a specified YouTube video.

B) FINAL TASK OUTPUT: A .zip file containing .wav audio samples extracted from the specified YouTube video. The samples should include vocals and instrumentals from key parts of the song (intro buildup, chorus/hook, outro). Samples should be chopped into 1/2/4/8 beat lengths based on the song's detected BPM. The .zip file should be uploaded to a server.

C) COMPONENTS OVERVIEW:
1. Download the audio from the specified YouTube video 
2. Extract the vocal and instrumental stems from the audio
3. Analyze the audio to detect BPM and identify key sections (intro, chorus, outro)  
4. Chop the vocal and instrumental audio from the key sections into sample lengths
5. Filter and select the best samples
6. Export the final audio samples in .wav format
7. Upload the .wav samples to a server in a .zip file

D) OVERVIEW OF SUBAGENTS:
- YouTube Download Agent: Downloads the audio from the specified YouTube video
- Audio Stem Separation Agent: Extracts the vocal and instrumental stems 
- Audio Analysis Agent: Detects BPM and identifies the intro, chorus and outro sections
- Audio Chopping Agent: Chops the key audio sections into 1/2/4/8 beat sample lengths  
- Sample Selection Agent: Filters and selects the best samples
- Audio Export Agent: Exports the final samples in .wav format
- File Upload Agent: Zips the .wav samples and uploads to a server

E) SUBAGENT 1: YouTube Download Agent
- Output: An audio file containing the audio from the specified YouTube video
- Tasks: Use the YouTube Data API to retrieve the video info, then use pytube library to download the audio stream, finally use pydub to convert the audio to a common format like .mp3
- Sequence: Retrieve YouTube video info > Download audio stream > Convert audio format > Output audio file

F) SUBAGENT 2: Audio Stem Separation Agent  
- Output: Separate .mp3 files for the vocal and instrumental stems
- Tasks: Use the Spleeter library to perform source separation on the audio file, extracting the vocal and accompaniment stems
- Sequence: Load audio file > Perform stem separation with Spleeter > Output vocal and instrumental .mp3 files

G) SUBAGENT 3: Audio Analysis Agent
- Output: Detected BPM and timestamps of the intro, chorus and outro sections
- Tasks: Use the librosa library to estimate BPM, loudness and onset strength. Identify sections by detecting significant changes in those features. 
- Sequence: Load audio files > Estimate BPM > Detect loudness and onsets > Identify intro, chorus, outro timestamps > Output BPM and section timestamps

H) SUBAGENT 4: Audio Chopping Agent
- Output: Folders of 1/2/4/8 beat length .wav samples for each section's vocals and instrumental 
- Tasks: Use pydub to slice the vocal and instrumental audio of each section into samples of the desired beat lengths based on the calculated BPM
- Sequence: Load section audio and BPM > Calculate sample lengths > Slice audio into samples > Output sample .wav files to folders

I) SUBAGENT 5: Sample Selection Agent
- Output: Curated selections of the best samples from each section/stem/sample length
- Tasks: Analyze each sample's spectral features, perceptual loudness, and novelty compared to other samples in its group. Use these metrics to rank and select the top samples.
- Sequence: Load all samples > Extract audio features > Rank samples by quality metrics > Select top samples > Output curated sample selections

J) SUBAGENT 6: Audio Export Agent  
- Output: High-quality, normalized .wav files of the final selected samples, named with a consistent convention
- Tasks: Use pydub to load the selected samples, normalize volume, convert to 16-bit PCM .wav format, and export with informative names
- Sequence: Load selected samples > Normalize volume > Convert to .wav > Generate informative file names > Export final .wav samples

K) SUBAGENT 7: File Upload Agent
- Output: A .zip file containing all the final .wav samples, uploaded to the specified server location 
- Tasks: Use the zipfile library to compress the .wav samples into a .zip archive. Use ftplib

subagent-summary-amended-for-actual-tasks

Here is the updated list of subagents and the specific tasks they would use to complete each component, based on the available tasks:

A) FULL TASK SUMMARY: Extract a collection of music samples in .wav format from a specified YouTube video.

E) SUBAGENT 1: YouTube Download Agent
- Output: An audio file containing the audio from the specified YouTube video
- Tasks: 
  - 43) Download youtube video to mp4
  - 44) Convert mp4 to mp3
- Sequence: Download YouTube video to server as mp4 > Convert mp4 to mp3 > Output mp3 URL

F) SUBAGENT 2: Audio Stem Separation Agent  
- Output: Separate .mp3 files for the vocal and instrumental stems
- Tasks:
  - 36) Extract instrumental/vocal stems
- Sequence: Load mp3 URL > Extract vocal and instrumental stems > Output vocal and instrumental mp3 URLs

G) SUBAGENT 3: Audio Analysis Agent
- Output: Detected BPM and timestamps of the intro, chorus and outro sections
- Tasks: 
  - 33) Get tempo (BPM) of mp3
  - 5) Extract visual data from image - REQUIRED to analyze spectrogram of mp3 to identify sections
- Sequence: Load vocal and instrumental mp3s > Get BPM of each > Generate spectrogram images > Analyze spectrograms to identify intro, chorus, outro timestamps > Output BPM and section timestamps

H) SUBAGENT 4: Audio Chopping Agent
- Output: Folders of 1/2/4/8 beat length .wav samples for each section's vocals and instrumental 
- Tasks:  
  - 45) Split/extract video by interval - REQUIRED, but for mp3 instead of mp4
  - REQUIRED: A task to convert mp3 samples to wav format
- Sequence: Load section vocal and instrumental mp3s > Calculate sample lengths from BPM > Extract samples as mp3 using split/extract task > Convert sample mp3s to wavs > Output folders of wav samples

I) SUBAGENT 5: Sample Selection Agent
- Output: Curated selections of the best samples from each section/stem/sample length
- Tasks:
  - 1) GPT4 - to analyze and rank samples based on audio features
- Sequence: Load all wav samples > Extract audio features with a library (REQUIRED) > Use GPT4 to rank samples by quality > Select top samples > Output curated sample selections

J) SUBAGENT 6: Audio Export Agent 
- Output: High-quality, normalized .wav files of the final selected samples
- Tasks:  
  - 9) Save content to server - to save the final wavs
  - REQUIRED: A task to normalize wav volume/quality   
- Sequence: Load selected wav samples > Normalize wav volume/quality > Save final wavs to server > Output URLs of final wav samples

K) SUBAG

claude 3 - mermaid-for-all

claude 3 - subagent1

subagent1-refined

Here is the full detailed breakdown for the YouTube Download Subagent:

A) SUBAGENT SUMMARY: Download the audio from a specified YouTube video as an MP3 file.

B) FINAL TASK OUTPUT: An MP3 audio file containing the audio from the specified YouTube video, outputted as a URL on the internal server.

C) SUBAGENT INPUT: The URL of the YouTube video to download the audio from (eg https://www.youtube.com/watch?v=dQw4w9WgXcQ)

D) SUBAGENT TASK SUMMARY:

Input: YouTube video URL
1. Download youtube video to mp4 using task 43
   - Input: YouTube video URL
   - Output: URL of downloaded MP4 video on internal server  
2. Convert downloaded mp4 to mp3 using task 44
   - Input: URL of MP4 video on internal server from step 1
   - Output: URL of converted MP3 audio on internal server
Output: URL of final MP3 audio file on internal server

E) SILOS: This subagent does not require silos as it is a simple linear 2-step process for a single video. Silos would only be needed if processing multiple videos.

F) JSON SCHEMA:

```json
{
  "subagent": "YouTube Download Agent",
  "input": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "steps": [
    {
      "task": 43,
      "input": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
      "output": "_server/downloaded_videos/dQw4w9WgXcQ.mp4"
    },
    {
      "task": 44,  
      "input": "_server/downloaded_videos/dQw4w9WgXcQ.mp4",
      "output": "_server/converted_audio/dQw4w9WgXcQ.mp3"
    }
  ],
  "output": "_server/converted_audio/dQw4w9WgXcQ.mp3"
}
```

random-autocomplete

To achieve the end goal of generating a collection of music samples from a YouTube video in a zip file, the most relevant tasks might be:

1. Download the audio from the YouTube video
2. Separate the audio into instrumental and vocal stems
3. Analyze the audio to determine the tempo and beat positions
4. Generate a spectrogram of the audio to visually identify key sections (intro, chorus, outro)
5. Extract 16-beat samples from the intro, chorus, and outro sections for both the instrumental and vocal stems
6. Chop up the extracted samples into 1, 2, 4, and 8 beat lengths based on the tempo
7. Filter the samples (if necessary) to reduce the total number
8. Save all the sample files to a server
9. Compress the samples into a zip file

The overall sequence would be:

1. Use a YouTube audio downloader API or library to extract the audio from the provided YouTube video link and save it as an audio file.

2. Utilize a stem separation library or API (e.g., Spleeter) to separate the audio into instrumental and vocal stems.

3. Analyze the audio using a library like librosa to determine the tempo (BPM) and beat positions.

4. Generate a spectrogram of the audio using a library like matplotlib to visually identify the intro, chorus, and outro sections.

5. Based on the spectrogram, extract 16-beat samples from the intro, chorus, and outro sections for both the instrumental and vocal stems using librosa or pydub.

6. Calculate the duration of each beat based on the tempo (e.g., at 90 BPM, 1 beat = 0.66 seconds) and use a library like pydub to chop the extracted samples into 1, 2, 4, and 8 beat lengths.

7. If the number of samples is too large, implement a filtering method to reduce the total number of samples based on specific criteria (e.g., remove similar samples, keep only the most representative samples from each section).

8. Save all the sample files to a designated folder on the server using Python's built-in file handling capabilities.

9. Use a library like zipfile to compress the folder containing the samples into a zip file.

The agent can accomplish this task by utilizing various APIs and Python libraries to perform audio downloading, stem separation, audio analysis, sample extraction, and file compression. The Windows 11 Virtual Machine can be used to run the necessary scripts and store the generated files.

subagent1refined-repeating

Here is the updated subagent with the full list of tasks in JSON format:

--------------
[subagent1refined]
--------------

A) SUBAGENT SUMMARY: Extract the vocal and instrumental stems from a YouTube video, analyze the spectrogram, and select key sections to generate audio samples.

B) FINAL TASK OUTPUT:
- Vocal stem audio file (.wav)
- Instrumental stem audio file (.wav) 
- Spectrogram image file (.png)
- 3 x 16-beat audio clips from the vocal stem (.wav) 
- 3 x 16-beat audio clips from the instrumental stem (.wav)

C) SUBAGENT INPUT: 
- YouTube video URL

E) SUBAGENT TASK SUMMARY:
YouTube URL input > Extract audio from video > Split audio into vocal and instrumental stems > Generate spectrogram of full audio > Select 3 key 16-beat sections (intro, chorus, outro) > Extract the 3 sections from both the vocal and instrumental stems > Output vocal stem, instrumental stem, spectrogram, and 6 audio clips

G) FULL TASKS (INCLUDING REPEATING TASKS):
```json
[
  {
    "task": "Extract audio from YouTube video",
    "input": "YouTube video URL",
    "output": "Full audio file"
  },
  {
    "task": "Split audio into vocal and instrumental stems",
    "input": "Full audio file",
    "output": [
      "Vocal stem audio file",
      "Instrumental stem audio file"
    ]
  },
  {
    "task": "Generate spectrogram of full audio",
    "input": "Full audio file", 
    "output": "Spectrogram image file"
  },
  {
    "task": "Select intro 16-beat section",
    "input": [
      "Vocal stem audio file",
      "Instrumental stem audio file",
      "Spectrogram image file"
    ],
    "output": [
      "Vocal intro 16-beat audio clip",
      "Instrumental intro 16-beat audio clip"  
    ]
  },
  {
    "task": "Select chorus 16-beat section", 
    "input": [
      "Vocal stem audio file",
      "Instrumental stem audio file",
      "Spectrogram image file"
    ],
    "output": [
      "Vocal chorus 16-beat audio clip",
      "Instrumental chorus 16-beat audio clip"
    ]
  },
  {
    "task": "Select outro 16-beat section",
    "input": [
      "Vocal stem audio file", 
      "Instrumental stem audio file",
      "Spectrogram image file"  
    ],
    "output": [
      "Vocal outro 16-beat audio clip",
      "Instrumental outro 16-beat audio clip"
    ]
  }
]
```

task-ID-numbers

subagent1-output-variables

Here is the updated subagent with the output tokens added:

--------------
[subagent1refined]
--------------

A) SUBAGENT SUMMARY: Extract the vocal and instrumental stems from a YouTube video, analyze the spectrogram, and select key sections to generate audio samples.

B) FINAL TASK OUTPUT:
- Vocal stem audio file (.wav) [vocal-stem]
- Instrumental stem audio file (.wav) [instrumental-stem]
- Spectrogram image file (.png) [spectrogram]
- 3 x 16-beat audio clips from the vocal stem (.wav) [vocal-intro-clip], [vocal-chorus-clip], [vocal-outro-clip]
- 3 x 16-beat audio clips from the instrumental stem (.wav) [instrumental-intro-clip], [instrumental-chorus-clip], [instrumental-outro-clip]

C) SUBAGENT INPUT: 
- YouTube video URL [youtube-url]

E) SUBAGENT TASK SUMMARY:
[youtube-url] input > Extract audio from video > Split audio into [vocal-stem] and [instrumental-stem] > Generate [spectrogram] of full audio > Select 3 key 16-beat sections (intro, chorus, outro) > Extract the 3 sections from both the [vocal-stem] and [instrumental-stem] > Output [vocal-stem], [instrumental-stem], [spectrogram], and 6 audio clips

G) FULL TASKS (INCLUDING REPEATING TASKS):
```json
[
  {
    "task": "Extract audio from YouTube video",
    "input": "[youtube-url]",
    "output": "[full-audio]"
  },
  {
    "task": "Split audio into vocal and instrumental stems",
    "input": "[full-audio]",
    "output": [
      "[vocal-stem]",
      "[instrumental-stem]"
    ]
  },
  {
    "task": "Generate spectrogram of full audio",
    "input": "[full-audio]", 
    "output": "[spectrogram]"
  },
  {
    "task": "Select intro 16-beat section",
    "input": [
      "[vocal-stem]",
      "[instrumental-stem]",
      "[spectrogram]"
    ],
    "output": [
      "[vocal-intro-clip]",
      "[instrumental-intro-clip]"  
    ]
  },
  {
    "task": "Select chorus 16-beat section", 
    "input": [
      "[vocal-stem]",
      "[instrumental-stem]",
      "[spectrogram]"
    ],
    "output": [
      "[vocal-chorus-clip]",
      "[instrumental-chorus-clip]"
    ]
  },
  {
    "task": "Select outro 16-beat section",
    "input": [
      "[vocal-stem]", 
      "[instrumental-stem]",
      "[spectrogram]"  
    ],
    "output": [
      "[vocal-outro-clip]",
      "[instrumental-outro-clip]"
    ]
  }
]
```

H) OUTPUT TOKENS:
- [full-audio]
- [vocal-stem]
- [instrumental-stem] 
- [spectrogram]
- [vocal-intro-clip]
- [instrumental-intro-clip]
- [vocal-chorus-clip] 
- [instrumental-chorus-clip]
- [vocal-outro-clip]
- [instrumental-outro-clip]
--------------

subagent1refined-repeating-json-core

Here is the updated subagent with the TASKFLOW SUMMARY section added:

--------------
[subagent1refined]
--------------

A) SUBAGENT SUMMARY: Extract the vocal and instrumental stems from a YouTube video, analyze the spectrogram, and select key sections to generate audio samples.

C) SUBAGENT INPUT: 
- YouTube video URL

G) TASKFLOW SUMMARY:
```
Title task: Extract audio from YouTube video
Description of the task: Download the audio from the provided YouTube video URL and save it as a high-quality audio file.
Input(s) involved: YouTube video URL
Output token name: Full audio file

Title task: Split audio into vocal and instrumental stems
Description of the task: Use source separation techniques to split the full audio file into separate vocal and instrumental stems.
Input(s) involved: Full audio file
Output token names: Vocal stem audio file, Instrumental stem audio file

Title task: Generate spectrogram of full audio
Description of the task: Analyze the full audio file and generate a spectrogram image that visualizes the frequency content over time.
Input(s) involved: Full audio file
Output token name: Spectrogram image file

Title task: Select intro 16-beat section
Description of the task: Using the spectrogram and audio stems, identify and extract the first 16 beats of the song for both the vocal and instrumental.
Input(s) involved: Vocal stem audio file, Instrumental stem audio file, Spectrogram image file
Output token names: Vocal intro 16-beat audio clip, Instrumental intro 16-beat audio clip

Title task: Select chorus 16-beat section
Description of the task: Using the spectrogram and audio stems, identify and extract a representative 16-beat section from the chorus for both the vocal and instrumental.
Input(s) involved: Vocal stem audio file, Instrumental stem audio file, Spectrogram image file
Output token names: Vocal chorus 16-beat audio clip, Instrumental chorus 16-beat audio clip

Title task: Select outro 16-beat section  
Description of the task: Using the spectrogram and audio stems, identify and extract the last 16 beats of the song for both the vocal and instrumental.
Input(s) involved: Vocal stem audio file, Instrumental stem audio file, Spectrogram image file
Output token names: Vocal outro 16-beat audio clip, Instrumental outro 16-beat audio clip
```
--------------

subagent1mermaid

<!DOCTYPE html>
<html>
<head>
  <title>Mermaid Diagram Example</title>
  <script src="https://cdn.jsdelivr.net/npm/mermaid/dist/mermaid.min.js"></script>
</head>
<body>
  <div class="mermaid">
graph TD
    A[YouTube video URL] --> B[Extract audio from YouTube video]
    B --> C[Full audio file]
    C --> D[Split audio into vocal and instrumental stems]
    D --> E[Vocal stem audio file]
    D --> F[Instrumental stem audio file]
    C --> G[Generate spectrogram of full audio]
    G --> H[Spectrogram image file]
    E --> I[Select intro 16-beat section]
    F --> I
    H --> I
    I --> J[Vocal intro 16-beat audio clip]
    I --> K[Instrumental intro 16-beat audio clip]
    E --> L[Select chorus 16-beat section]
    F --> L
    H --> L
    L --> M[Vocal chorus 16-beat audio clip]
    L --> N[Instrumental chorus 16-beat audio clip]
    E --> O[Select outro 16-beat section]
    F --> O
    H --> O
    O --> P[Vocal outro 16-beat audio clip]
    O --> Q[Instrumental outro 16-beat audio clip]
  </div>

prompts-required

I apologize for the confusion, but I don't have enough information yet to provide the list of prompts you requested. You mentioned an "above list of components" but that list is not present in our conversation so far.

To summarize where we're at:

1. You described a high-level task of extracting music samples from YouTube videos.

2. You expressed wanting to know the final output/end goal before starting. However, you haven't provided your description of what that final output should be.

3. You mentioned breaking the task down into component sub-tasks. However, you haven't provided the actual list of component tasks.

Without knowing the details of the desired final output and the breakdown of component tasks, I don't have enough context to suggest the specific prompts that would be needed at each step.

If you can provide 1) your description of the ideal final output, and 2) the breakdown of component tasks, I'll be happy to review those and suggest what unique prompts may be needed for each component task to achieve the final goal. Just include those missing pieces and I'll take another look!

Let me know if you have any other questions!