{
  "id": 162050,
  "title": "Spectogram",
  "url": "/competitions/birdsong-recognition/discussion/162050",
  "author_name": "",
  "post_date": "2020-06-27T06:03:20.348842800Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>spectogram shows the magnitude (change in frequency) against time and frequency a 3D plot, my question here while looking at spectogram many notebooks mentioned that </p>\n\n<p>\"since we see that all action is taking place at the bottom of the spectrum, we can convert the frequency axis to a logarithmic one\"</p>\n\n<p>I am lost what does mean by above statement? As i know we use log for constant effect of magnitude, is it that reason we convert to log for constant effect in spectogram if that is the case then what above quoted statement explaining about?</p>",
  "messages": [
    {
      "id": "903818",
      "postDate": "06/27/2020 06:03:20",
      "content": "<p>spectogram shows the magnitude (change in frequency) against time and frequency a 3D plot, my question here while looking at spectogram many notebooks mentioned that </p>\n\n<p>\"since we see that all action is taking place at the bottom of the spectrum, we can convert the frequency axis to a logarithmic one\"</p>\n\n<p>I am lost what does mean by above statement? As i know we use log for constant effect of magnitude, is it that reason we convert to log for constant effect in spectogram if that is the case then what above quoted statement explaining about?</p>",
      "rawMarkdown": "spectogram shows the magnitude (change in frequency) against time and frequency a 3D plot, my question here while looking at spectogram many notebooks mentioned that \n\n\"since we see that all action is taking place at the bottom of the spectrum, we can convert the frequency axis to a logarithmic one\"\n\nI am lost what does mean by above statement? As i know we use log for constant effect of magnitude, is it that reason we convert to log for constant effect in spectogram if that is the case then what above quoted statement explaining about?",
      "votes": null
    },
    {
      "id": "904384",
      "postDate": "06/27/2020 15:27:26",
      "content": "<p>Studies have shown that humans are better at distinguishing between lower frequencies compared with higher ones. For example, our ears can easily tell the difference between 110Hz(A) and 116Hz(A#). On the other hand, It's hard for us to tell apart 5100Hz and 5106Hz even though both pairs of frequencies are spaced by 5Hz.</p>\n\n<p>The Mel scale is an approximate model for how the human ears perceive frequency/pitch. Mathematically, there is a logarithmic relationship between the actual frequency vs the mel scale.</p>\n\n<p>In python, you can use the librosa.feature.melspectrogram to give you the frequency mapping using the mel scale.</p>",
      "rawMarkdown": "Studies have shown that humans are better at distinguishing between lower frequencies compared with higher ones. For example, our ears can easily tell the difference between 110Hz(A) and 116Hz(A#). On the other hand, It's hard for us to tell apart 5100Hz and 5106Hz even though both pairs of frequencies are spaced by 5Hz.\n\nThe Mel scale is an approximate model for how the human ears perceive frequency/pitch. Mathematically, there is a logarithmic relationship between the actual frequency vs the mel scale.\n\nIn python, you can use the librosa.feature.melspectrogram to give you the frequency mapping using the mel scale.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 904384,
      "author_name": "sayujnath",
      "author_url": "",
      "post_date": "06/27/2020 15:27:26",
      "content": "<p>Studies have shown that humans are better at distinguishing between lower frequencies compared with higher ones. For example, our ears can easily tell the difference between 110Hz(A) and 116Hz(A#). On the other hand, It's hard for us to tell apart 5100Hz and 5106Hz even though both pairs of frequencies are spaced by 5Hz.</p>\n\n<p>The Mel scale is an approximate model for how the human ears perceive frequency/pitch. Mathematically, there is a logarithmic relationship between the actual frequency vs the mel scale.</p>\n\n<p>In python, you can use the librosa.feature.melspectrogram to give you the frequency mapping using the mel scale.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "903818": "spectogram shows the magnitude (change in frequency) against time and frequency a 3D plot, my question here while looking at spectogram many notebooks mentioned that \n\n\"since we see that all action is taking place at the bottom of the spectrum, we can convert the frequency axis to a logarithmic one\"\n\nI am lost what does mean by above statement? As i know we use log for constant effect of magnitude, is it that reason we convert to log for constant effect in spectogram if that is the case then what above quoted statement explaining about?",
    "904384": "Studies have shown that humans are better at distinguishing between lower frequencies compared with higher ones. For example, our ears can easily tell the difference between 110Hz(A) and 116Hz(A#). On the other hand, It's hard for us to tell apart 5100Hz and 5106Hz even though both pairs of frequencies are spaced by 5Hz.\n\nThe Mel scale is an approximate model for how the human ears perceive frequency/pitch. Mathematically, there is a logarithmic relationship between the actual frequency vs the mel scale.\n\nIn python, you can use the librosa.feature.melspectrogram to give you the frequency mapping using the mel scale."
  },
  "source": "meta"
}