Using audiosamples/volume of AudioSource to simply animate a Jaw Transform. As seen here: https://youtu.be/5P_mWSFA4ow

JawAnimator.cs

[SerializeField] private AudioSource _linkedAudioSource = null;
[SerializeField] private float _minRmsValue = 0.0001f; // minimum to consider "talking"
[SerializeField] private Renderer _rendererToCheckVisibility = null;
[SerializeField] private float _jawTalkSpeed = 0.12f;

private JawStates _jawState = JawStates.Default;
private float[] _samples = new float[512]; // For spectrum data
private float _rmsValue; // Root mean square, representing audio level

public enum JawStates
{
	Default,
	Talking
}

void GetAudioData()
{
	_linkedAudioSource.GetSpectrumData(_samples, 0, FFTWindow.BlackmanHarris);
}

void CalculateRMS()
{
	float sum = 0;
	for (int i = 0; i < _samples.Length; i++)
	{
		sum += _samples[i] * _samples[i];
	}

	decimal decimalValue = Math.Round((decimal)Mathf.Sqrt(sum / _samples.Length), 4);
	_rmsValue = (float)decimalValue;
}

void Update()
{
	bool RmsIsLoudEnough()
	{
		GetAudioData();
		CalculateRMS();
		return rmsValue >= minRmsValue;
	}

	if (Time.frameCount % 10 == 0) // only check once every 10 frames
	{
		// this if condition is complex because we don't want to calculate things when not necessary. the most expensive calculation is last in the condition
		if (_linkedAudioSource.clip != null && _linkedAudioSource.isPlaying && _rendererToCheckVisibility.isVisible && RmsIsLoudEnough())
		{
			if (_jawState != JawStates.Talking)
				JawTalking();
		}
		else
		{
			if (_jawState != JawStates.Default)
				JawShut();
		}
	}
}

void JawTalking()
{
	jawState = JawStates.Talking;
	// Insert your "jaw is now talking" animation/tweening logic here. I did a simple pingponging tween of the LocalRotation of the jaw transform.
}

void JawShut()
{
	jawState = JawStates.Default;
	// Insert your "jaw shuts" animation/tween logic here. I did a simple one-time quick tween of the LocalRotation of the jaw transform to shut and stay shut.
}