2016-05-03 To 2016-05-09

来自cslt Wiki
2016年5月20日 (五) 01:46Linjy讨论 | 贡献的版本

(差异) ←上一版本 | 最后版本 (差异) | 下一版本→ (差异)
跳转至: 导航搜索

20160504

Confirm 'voice conversion' as my major direction. Get several literature and tool kits to be the studying source for the next two weeks. Learn how to use 'putty' and get some message about how to study and work in our lab. Operate the 'gmm_vc' successfully. 20160506

Read literature about 'Exemplar-Based Sparse Representation with Residual Compensation for Voice Conversion'. Read literature about 'Sparse representation for frequency warping based voice conversion'. Read literature about 'Correlation-based frequency warping for voice conversion'.

20160507

Read literature about 'Voice Conversion Based on Maximum-Likelihood Estimation of Spectral Parameter Trajectory'. Run and understand the first part and the second part of 'gmm_vc' toolkit.

20160508

Run and understand the third part of the 'gmm_vc' toolkit. Understand how a wave file can be 'analysized' to mcc and pitch, and how mcc and pitch can also be used to synthesize the original wav. Run the toolkit of sptk3.8 to implement the transform from mcc to wav. Try to finish the whole conversion process.