We Recommend Headphone Listening. Always!
Time-scale Modification (TSM) Demo
Our proposed method is μTVS. The other methods are shown for comparison.
The details of the method can be learnt from the paper
Mel-scale sub-band modelling for perceptually improved time-scale modification of speech and audio signals,
published at the National Conference on Communications 2017, IEEE.
Examples of time-scaled outputs is presented below.
Scaling factor of α=0.5
| Name | Original | WSOLA | PV | PV-IP | HPS | μTVS |
|---|---|---|---|---|---|---|
| Fspeech | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Mspeech | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download | s
|
| Flute | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Glocs | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Jazz | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Tsynth | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Vocals | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
Scaling factor of α=1.5
| Name | Original | WSOLA | PV | PV-IP | HPS | μTVS |
|---|---|---|---|---|---|---|
| Fspeech | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Mspeech | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Flute | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Glocs | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Jazz | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Tsynth | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
|
| Vocals | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
Screenshot of GUI used in evaluating the above sound samples using Listening Test
In the screenshot: A,B,...,G are the 7 different sound kinds. The bottom pushbottons - 1,2,3,4,5 carry the signal from one of the 5 algorithms (randomly ordered).
Pitch-scale Modification Demo
| Name | Original | β=1.35 | β=0.85 |
|---|---|---|---|
| Speech(M) | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
| Vocals(F) | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
| Vocals(M) | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
Time-varying Time-scale Modification Demo
| Name | Speech(M) | Vocals(F) | Vocals(M) |
|---|---|---|---|
| Original | ▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |
| Non-uniform TSM |
▶
▮▮
Download |
▶
▮▮
Download |
▶
▮▮
Download |