Linear predictive coding of audio

Original audio: Mono, 16 bits/sample, 44100 samples/s, 705.6 kbps

Uniform quantization 4 levels. 2 bits/sample (88.2 kbps) without source coding.

• No predictor. 9.7 dB, Huffman 2 bits/sample (88.2 kbps).
• Predictor order 1. 18.9 dB, Huffman 1.69 bits/sample (74.5 kbps).
• Predictor order 2. 18.6 dB
• Predictor order 3. 20.1 dB
• Predictor order 4. 19.4 dB
• Predictor order 5. 18.8 dB, Huffman 2 bits/sample (88.2 kbps).
• Predictor order 6. 18.8 dB
• Predictor order 7. 19.0 dB
• Predictor order 8. 19.1 dB, Huffman 2 bits/sample (88.2 kbps).

Uniform quantization 32 levels. 5 bits/sample (220.5 kbps) without source coding.

• No predictor. 26.1 dB, Huffman 4.70 bits/sample (207.3 kbps).
• Predictor order 1. 29.7 dB, Huffman 2.70 bits/sample (119.1 kbps).
• Predictor order 2. 29.5 dB
• Predictor order 3. 31.4 dB
• Predictor order 4. 31.6 dB
• Predictor order 5. 32.9 dB, Huffman 2.87 bits/sample (126.6 kbps).
• Predictor order 6. 32.9 dB
• Predictor order 7. 32.6 dB
• Predictor order 8. 32.6 dB, Huffman 2.90 bits/sample (127.9 kbps).

Uniform quantization 256 levels. 8 bits/sample (352.8 kbps) without source coding.

• No predictor. 42.8 dB, Huffman 7.42 bits/sample (327.2 kbps).
• Predictor order 1. 45.1 dB, Huffman 5.00 bits/sample (220.5 kbps).
• Predictor order 2. 45.1 dB
• Predictor order 3. 47.8 dB
• Predictor order 4. 47.7 dB
• Predictor order 5. 48.5 dB, Huffman 4.87 bits/sample (214.8 kbps).
• Predictor order 6. 48.5 dB
• Predictor order 7. 48.1 dB
• Predictor order 8. 48.0 dB, Huffman 4.77 bits/sample (210.4 kbps).

Lossless coding with FLAC. Original files in stereo, 16 bits/sample, 44100 samples/s